Agomab Therapeutics reported the 48-week results from the open-label extension of its Phase 2a STENOVA trial on October 1, and the number the company led with is a count of medical procedures that did not happen. Across 49 patients with fibrostenosing Crohn's disease who took ontunisertib for up to 60 weeks, one needed an endoscopic balloon dilation and none needed surgery. Agomab converted that into an annualized event rate of 3 percent and placed it beside the 23 to 35 percent rate that recent publications report for comparable patients on advanced therapies.
That comparison is the most arresting claim in the release and the one most in need of a caveat, which Agomab supplied itself in a footnote: cross-trial comparisons are inherently limited by differences in design, population and endpoints, and no conclusions about relative efficacy should be drawn from them without head-to-head data. The more useful question is not whether ontunisertib beats the historical rate. It is what an open-label extension can carry into a randomized trial, and whether the trial that is opening next is designed to answer the question the extension could only raise.
A scar that no approved drug is aimed at
Crohn's disease is treated as an inflammatory condition, and the drugs that dominate its treatment, including the biologics many STENOVA patients were already taking, suppress that inflammation. The complication that sends people to the operating room is a different problem: a segment of bowel narrowed by fibrosis laid down over years of cycles, stiff and scarred enough that contents cannot pass. Florian Rieder of the Cleveland Clinic, who joined the company's webcast, described a disease with "no approved therapies targeting the fibrotic component."
What exists instead is mechanical. Endoscopic balloon dilation opens a stricture by force, and surgery removes the segment. Both are temporary answers to a process that keeps producing scar, and repeat procedures accumulate over a patient's life. Ontunisertib is an oral small molecule aimed at the biology underneath: it inhibits ALK5, also called TGF-beta receptor I, the receptor through which TGF-beta signals fibroblasts to lay down collagen. Targeting that pathway has been constrained for years by systemic toxicity, which is why a version designed to stay in the gut has drawn attention from a field that has watched this target fail elsewhere. The drug's design answer is to act only where the disease is. Rapid first-pass metabolism in the liver limits systemic exposure, and the company reported that plasma concentrations in the extension stayed well below the level needed to inhibit the target, the same gut-restricted profile seen in the randomized portion of the trial. The FDA granted it Fast Track designation.
What the extension measured, and what it could not
STENOVA ran in two parts. The first randomized 103 patients with symptomatic ileal strictures to 100 milligrams once daily, 200 milligrams twice daily, or placebo for 12 weeks on top of standard of care, including anti-inflammatory biologics, across sites in the United States, Canada and six European countries. The primary endpoint was safety and tolerability, with pharmacokinetics, target engagement and exploratory imaging as the follow-up measures.
The second part is the data released this week. Of the Part A patients eligible to continue, 94 percent did, 49 people, all of them moving to 200 milligrams twice daily for up to 48 additional weeks, with 37 completing. There was no placebo arm. Everyone in the extension was on the drug, and the patients who entered it were the ones who had finished the blinded phase, chosen to continue, and in many cases were already doing well enough to want to. The release's own description of the extension is careful about what it was for: long-term safety and pharmacokinetics, with the event rate and disease progression explored alongside them.
The 3 percent is a count of absent events
One dilation across 49 patients and up to 60 weeks of treatment is a small numerator with a large denominator of uncertainty around it. In a single-arm study, an event count is a statement about the patients who enrolled as much as about the drug. A group whose strictures were less severe at baseline will have fewer dilations whether the drug works or not, and the same is true of a group that has already passed through the first 12 weeks of a trial on treatment. The 23 to 35 percent figures come from publications about comparable populations on advanced therapies, and comparable is doing a great deal of work in that sentence: baseline severity, follow-up length and how events are counted all move a rate of this kind, which is why the company's own footnote warns against the comparison. A rate built from a single event is also fragile in a way the percentage hides. Move the one dilation a few weeks earlier or later, or count the patient-years slightly differently, and 3 percent becomes 5 or 2, with no confidence interval in a press release to say how wide the real range is.
None of that makes the number meaningless. A 3 percent annualized rate is low enough, across a population selected precisely because its strictures were symptomatic, that a placebo arm is now worth the cost of running one. The extension was not built to prove an effect on procedures, and the trial that is will be the first place the number can be tested rather than quoted.
Strictures that stopped narrowing, and the patients who caught up
The imaging findings are the part of the release with an internal comparison. Magnetic resonance enterography at weeks 24 and 48 of the extension suggested that strictures stabilized over treatment. During the 12-week blinded portion, patients on placebo had shown a trend of radiological progression; after they switched to ontunisertib in the extension, that progression stopped, and the release describes them progressively catching up with the patients who had been on the drug from the start.
A switcher pattern like that is worth more than a cross-trial rate, because both groups are inside the same study, treated the same way, and read by the same radiologists. It is also the earliest signal in the program that points at the fibrotic process itself rather than at symptoms. Two limits stay attached to it. Twelve weeks of placebo data is a short window for a scar that took years to form, so the progression trend it interrupts is a small sample of a slow process. And imaging reads in an open-label extension are unblinded, which matters more for a judgment like stabilization than for a measurement like a dilation.
The endoscopy results, meanwhile, could not be read at all. The week 48 endoscopy was optional, and the company said the data were too limited for a reliable interpretation. That gap is the one to keep in view, because endoscopic passability is the endpoint the next trial will use.
Symptoms held steady across the extension, with patients maintaining low disease activity on the Crohn's Disease Activity Index and quality-of-life measures, and 97 percent of evaluable patients in clinical remission by that index at week 48. Those are open-label numbers with no comparator, and a symptom index can improve for reasons other than the drug, which is the standard objection to reading too much into them.
The trial that will decide whether the mechanism works
NOV-ERA, the Phase 2b study Agomab has finished regulatory filings for, is a different kind of test. It will randomize about 320 adults with symptomatic fibrostenosing Crohn's disease, one to one to one to one, to 400, 200 or 100 milligrams twice daily or placebo, with a 52-week treatment period. To enroll, patients must have at least one naive or anastomotic ileal stricture that a central reader confirms cannot be passed endoscopically. The primary endpoint is the proportion of patients whose index stricture becomes passable at week 24, judged by that same kind of central read.
That design answers the two questions the extension left open. A placebo arm establishes what happens to unpassable strictures without the drug, which is the denominator the 3 percent has never had. An objective endoscopic endpoint, read centrally and reported as pass or fail, removes the judgment calls that a symptom index and an unblinded scan invite. Week 24 is earlier than anything in the extension, so the trial will also show whether the effect needs a year of dosing or begins to appear within six months. The 400 milligram arm is supported by Phase 1 tolerability at that dose and by the extension, where the drug's safety profile held across the longer exposure.
What the evidence can carry forward
The safety findings are the reason a larger trial is plausible. Five serious adverse events occurred in four patients during the extension, all judged unrelated or unlikely related to treatment, and the company reported no signals of cardiac injury, pro-inflammatory effects, vasculitis, liver toxicity or thrombotic events across the extended treatment. That list is not random. Those are the events a sponsor would look for when inhibiting this pathway, and the company's stated rationale is that keeping the drug out of the bloodstream buys a better safety profile than the systemically available inhibitors in the class. Their absence over up to 60 weeks of dosing is the finding that makes a gut-restricted version of the drug worth testing at scale. The tolerability held well enough that most patients who could continue did, and most of those finished.
Investors read the release as good news. Agomab shares rose 6.3 percent in premarket trading on October 1.
The argument for ontunisertib does not depend on the event rate being exactly 3 percent, and it should not be sold that way. The argument is that fibrosis is the part of Crohn's disease that approved drugs leave alone, that this molecule reaches the tissue where the scar forms without the systemic exposure that made the target difficult, and that the extension found no safety signal strong enough to stop testing the idea. An open-label study of 49 patients can establish that much and no more. The claim that patients will need fewer procedures is the one that requires a control arm, a central endoscopy read and a placebo group, and that is what NOV-ERA is built to provide. The next number in this program will come from patients who did not volunteer to keep taking the drug, read by endoscopists who do not know which arm they are in. That is the only way the count of procedures that did not happen becomes evidence a regulator can weigh.
Primary sources
- Agomab Therapeutics, press release furnished with the October 1, 2026 Form 6-K, for the STENOVA extension results, the event rate and its cross-trial caveat, the imaging findings, the safety summary, and the NOV-ERA design.
- ClinicalTrials.gov, STENOVA record, for the Phase 2a design, the randomization scheme and the rollover criteria.
- ClinicalTrials.gov, NOV-ERA record, for the Phase 2b design, eligibility and primary endpoint.
- El Ouali et al., UEG Journal, 2022, one of the references the company cites for the historical annualized event rates in patients on advanced therapies.
- Yahoo Finance, Agomab stock coverage, October 1, 2026, for the premarket share move.
- Clinical Trial Vanguard, STENOVA coverage, for the field's framing of the fibrosis-first thesis and the systemic toxicity constraint on the ALK5 class.