Balinatunfib in ulcerative colitis: an oral TNF inhibitor stopped at interim analysis

What was tried

Sanofi ran SPECIFI-UC (NCT06867094), a Phase 2, multinational, multicenter, randomized, double-blind, placebo-controlled dose-ranging study of balinatunfib in adults with moderate to severe ulcerative colitis. The design was parallel assignment with quadruple masking. Three oral tablet dose regimens were compared against matching placebo. Actual enrollment was 170 participants across 117 listed sites. The primary endpoint was the proportion of participants achieving clinical remission at Week 12 by modified Mayo Score, defined as a total score of 2 or less with no subscore above 1, a rectal bleeding subscore of 0 and a modified Mayo endoscopic subscore of 0 or 1 without friability.

Eligibility required adults aged 18 to 75 with endoscopically confirmed active disease, a modified Mayo Score of 5 to 9, an endoscopic subscore of at least 2 confirmed by central reader, and disease extent greater than 15 cm from the anal verge. Participants had to have failed, lost response to, or been intolerant of either conventional therapy or at least one approved advanced therapy. The study started on 28 March 2025, reached actual primary completion on 6 August 2026, completed on 12 October 2026, and was posted as terminated on 21 October 2026. No results have been posted. The registry whyStopped field reads "Business decision."

The biological hypothesis

Balinatunfib (ChEMBL6068043, a small molecule at ChEMBL max phase 2) was built to do with a pill what injected antibodies have done since infliximab: remove TNF-alpha from the inflamed gut. The compound belongs to a class of small molecules that bind within the TNF trimer and lock it into a distorted conformation, so that the ligand still assembles but no longer drives productive TNFR1 signaling (Vugler et al. 2022, Dömling 2025). Open Targets rates TNF tractable by small molecule with a high-quality pocket, a high-quality ligand and a solved structure with ligand bound, which is unusual for a soluble cytokine.

The disease-side rationale was strong on paper and narrow in type. Open Targets scores the TNF to ulcerative colitis association at 0.6336, built almost entirely from clinical evidence (0.9922) and literature (0.9887), with a weak RNA expression contribution (0.0247) and no genetic association score returned at all. The target is validated by the fact that anti-TNF biologics work, not by human genetics pointing at TNF in colitis.

What actually happened

Sanofi answered the question in its own words before the registry did. In the Q2 2026 results release of 30 July 2026, the company stated that interim analyses of the two ongoing Phase 2 studies in Crohn's disease (NCT06637631) and ulcerative colitis (NCT06867094) "showed that the studies did not meet the internal efficacy expectations on clinical endpoints and/or the primary endpoint of clinical remission," and that the development programme would be discontinued.

That was the third indication to read out below expectation. The Phase 2 psoriasis study SPECIFIC-PSO (NCT06073119) enrolled 221 patients and missed statistical significance on PASI-75 at 12 weeks in April 2025, with Sanofi then positioning the molecule for combination use rather than monotherapy. Rheumatoid arthritis, Crohn's disease and ulcerative colitis continued after that miss. All three have now stopped.

Safety was never the issue on the public record. openFDA returned no FAERS reports for balinatunfib, and the psoriasis readout was described as generally well tolerated across doses with no new safety concerns.

Failure mechanism, best guess

This is an efficacy failure wearing a business-decision label. The registry field and the sponsor's own press release disagree, and the press release is the more specific document.

The most parsimonious mechanism is an exposure ceiling rather than a wrong target. Anti-TNF antibodies achieve near-complete and sustained neutralization of a cytokine that is produced in bulk at the inflamed mucosa and exists in both soluble and membrane-bound forms. A tablet dosed to a safe systemic exposure has to hold that same pool suppressed between doses, in tissue, against local production. Directional activity in psoriasis with a sub-significant effect size is the signature of partial target engagement, not of a mistaken hypothesis.

A second contributor is population. The protocol admitted patients who had already failed an approved advanced therapy, a group that by construction is enriched for disease biology that does not depend on TNF. Recent work points the same way: a longitudinal single-cell atlas of anti-TNF treatment in inflammatory bowel disease mapped the cellular programmes that separate responders from non-responders (Thomas et al. 2024), and a 2026 preprint reports that a dysregulated stromal remodelling programme characterises prior anti-TNF failure in ulcerative colitis. If the non-responder state is stromal, a better TNF inhibitor does not reach it.

How to prevent this next time

No endpoint-level results were posted for this trial, and the sponsor released no remission rates, confidence intervals or dose-response data. There is nothing here to support a quantitative posterior, a power recalculation or a subgroup estimate, so the levers below are qualitative and are offered as such.

Biomarker and population enrichment is the first lever. The eligibility criteria allowed both advanced-therapy-naive and advanced-therapy-experienced participants into a 170-patient dose-ranging study with three active arms. Stratified randomisation by prior advanced therapy exposure, with a pre-specified naive-only analysis population, would have separated the exposure question from the biology question.

Base-rate adjustment is the second. By the time the UC study started in March 2025, the class had already produced a sub-significant monotherapy readout in psoriasis, a disease where anti-TNF biologics deliver large and reproducible effects. A programme-level prior built on that single fact argues for a pharmacodynamic gate before committing three indications in parallel.

Red-teaming is the third. The question worth asking was not whether oral TNF inhibition is possible but what fraction of antibody-level neutralization the tolerated dose achieves in inflamed tissue, and what remission rate that fraction predicts.

The single highest leverage change would have been to require a tissue-level target engagement readout from the psoriasis study, and to gate the Crohn's and ulcerative colitis starts on it rather than running all three indications in parallel.

What this means for similar programs

Claidex now holds three TNF post-mortems, and the recomputed Mechanism Risk Score for TNF is 68 of 100, in the orange band. The score is driven less by phase burden (19.0 of 30) than by archetype severity (19.4 of 25) and saturation (14.4 of 15), the latter reflecting 16 distinct programmes counted across the Claidex graph and the Open Targets drug list for this target.

The lesson generalises past TNF. When a target is validated by the success of one modality rather than by human genetics, a new modality against that target inherits none of the clinical validation and all of the competitive bar. The incumbent antibodies set the effect size a tablet must match, and partial engagement is not a partial win.

Open questions

What were the Week 12 remission rates by dose arm, and was there a dose-response trend that stopped short of significance? What mucosal exposures did the three regimens achieve, and what fraction of TNF neutralization did they correspond to? Did the advanced-therapy-naive subgroup behave differently from the experienced subgroup? Sanofi has said it is exploring combination use for this molecule, which will be interpretable only if the monotherapy exposure data are released.

Sources

1. • ClinicalTrials.gov, NCT06867094, SPECIFI-UC: https://clinicaltrials.gov/study/NCT06867094 - ClinicalTrials.gov, NCT06637631, Crohn's disease companion study: https://clinicaltrials.gov/study/NCT06637631 - Sanofi, Q2 2026 results press release, 30 July 2026, balinatunfib section: https://www.sanofi.com/assets/dotcom/pressreleases/2026/2026-07-30-05-30-00-3335767-en.pdf - Open Targets Platform, TNF (ENSG00000232810) and ulcerative colitis (MONDO_0005101), accessed 22 October 2026: https://platform.opentargets.org/target/ENSG00000232810 - ChEMBL, BALINATUNFIB, CHEMBL6068043: https://www.ebi.ac.uk/chembl/compound_report_card/CHEMBL6068043/ - Dömling A. Balinatunfib: A Clinical Oral Small Molecule TNFα Inhibitor. ChemMedChem, 2025. https://- Vugler A, et al. An orally available small molecule that targets soluble TNF to deliver anti-TNF biologic-like efficacy. Front Pharmacol, 2022. https://- Zhang N, et al. Selective inhibition of tumor necrosis factor receptor-1 (TNFR1) for the treatment of autoimmune diseases. Cytokine Growth Factor Rev, 2020. https://- Siegmund B, et al. TNF and TNF receptors as therapeutic targets for rheumatic diseases and beyond. Nat Rev Rheumatol, 2023. https://- Thomas T, et al. A longitudinal single-cell atlas of anti-tumour necrosis factor treatment in inflammatory bowel disease. Nat Immunol, 2024. https://- Veerasubramanian PJ, et al. Targeting TNF/TNFR superfamilies in immune-mediated inflammatory diseases. J Exp Med, 2024. https://- A dysregulated stromal remodelling programme characterises prior anti-TNF failure in ulcerative colitis. bioRxiv, 12 August 2026. https://- Clinical Trials Arena, Sanofi axes development of oral TNF inhibitor as monotherapy following Phase II fail, 24 April 2025: https://www.clinicaltrialsarena.com/news/balinatunfib-sanofi-monotherapy-fail/ - pharmaphorum, Sanofi drops two more programmes in R&D clear-out, 30 July 2026: https://pharmaphorum.com/news/sanofi-drops-two-more-programmes-rd-clear-out - openFDA drug event endpoint, queried for balinatunfib on 22 October 2026, no matching records: https://api.fda.gov/drug/event.json.