COVALENT-112: icovamenib (oral menin inhibitor) in type 1 diabetes

What was tested and why it matters

Menin is a protein inside beta cells that acts as a brake on their growth. Icovamenib (BMF-219) is a once-daily pill that binds menin covalently and releases that brake for a while. COVALENT-112 was the test of that idea in type 1 diabetes: could a short course of a pill get the body to rebuild some of its own insulin-producing capacity?¹

That question matters because almost everything else in the "cure" column involves importing cells — from a donor or a stem-cell line — with the surgery and immunosuppression that come with them. A pill that regrows what is already there would be a fundamentally easier proposition. Its small open-label dataset does not establish that new beta cells were produced in people.

Who was in it

Adults aged 18 to 70 with stage-3 T1D, at least one T1D autoantibody, an HbA1c of 6.5-10.0%, and — crucially — some insulin production left: entry required a C-peptide of at least 0.2 nmol/L for people diagnosed within three years, or at least 0.08 nmol/L for those diagnosed three to fifteen years ago.¹ The drug acts on beta cells that still exist, so people with none left were not eligible.

Thirty-seven people enrolled across 11 sites (nine in the United States, two in Canada). Dosing was 100 mg or 200 mg once daily for 12 weeks, followed by 40 drug-free weeks of observation. The primary endpoint was the mean change from baseline in stimulated C-peptide AUC at 26 weeks.¹

This trial is closed. The registry now lists it as completed, following an update posted 28 August 2026. The previous refresh recorded termination for portfolio prioritization. A registry status correction does not strengthen the uncontrolled efficacy evidence.¹

What it found

In the cohort diagnosed within the previous three years and dosed at 200 mg, mean stimulated C-peptide AUC was about 52% above baseline at Week 12.² Then the drug stopped. Forty weeks later, at Week 52, mean C-peptide AUC was still only about 7% below baseline, against the roughly 47% annual decline reported in historical placebo cohorts.³ The 100 mg arm showed less effect, and a longer-diagnosed cohort (three to fifteen years) broadly held its C-peptide.²

At the ADA's 86th Scientific Sessions (5-8 June 2026), Biomea added translational data: cytokine profiling showed no evidence of systemic immune activation, with inflammatory markers stable or reduced through Week 52.³

One thing that gets misreported: an HbA1c benefit was announced at those same sessions — a 1.2% reduction versus a 0.6% placebo increase — but that came from COVALENT-111 in type 2 diabetes, in people on background GLP-1 therapy.³ No numeric HbA1c result was detailed for this T1D trial. Type 2 evidence is not type 1 evidence.

How much weight to put on it

Less than the headline suggests, and here is exactly why.

The 200 mg cohort was five people, as reported by Biomea. The 100 mg cohort was six; the longer-diagnosed cohort, nine.² The reported data are open-label — everyone knew what they were taking. The randomized, placebo-controlled Part 2 was never completed, so the data reported by Biomea should be interpreted as open-label proof of concept, not a completed placebo-controlled comparison.¹² The 47% comparator is a historical figure from other studies, not a control arm in this one.

Every number above comes from company press releases and a conference presentation. No results are posted to ClinicalTrials.gov, and there is no peer-reviewed publication.¹ That is the weakest evidence grade this site records for a human efficacy claim.

There is also a safety history. On 6 June 2024 the FDA placed a full clinical hold on the whole BMF-219 diabetes programme, prompted by possible drug-induced hepatotoxicity — liver toxicity — observed during dose escalation in COVALENT-111, the type 2 diabetes study.⁴ The hold was lifted on 26 September 2024.⁵ Lifting the hold permitted development to resume; it did not establish that liver risk had disappeared. Larger controlled studies are needed to characterize safety.

What would settle it

A randomized, placebo-controlled trial, properly sized, with the results published rather than announced. Biomea has said it plans a larger Phase 2 in recently diagnosed T1D with longer dosing and a combination arm pairing icovamenib with an immunosuppressive agent — an acknowledgement that regrowing beta cells does nothing if the immune attack that destroyed the first set is still running. A future trial plan is not evidence that recruitment or dosing has begun.