Beta-cell regeneration / regrowth agents

Beta-cell regeneration asks a different question than transplantation: instead of importing cells from a donor or a stem-cell line, can we get the body to rebuild its own insulin-producing capacity? Most people with T1D are reported to retain a small number of beta cells even years after diagnosis, so the goal is to wake those cells up and multiply them — or to convert neighbouring cell types into insulin-producers.¹

The leading approach: DYRK1A inhibitors

Adult human beta cells almost never divide. In 2015 a Mount Sinai team screened thousands of compounds and found that harmine — a natural plant alkaloid — pushes them back into the cell cycle by blocking an enzyme called DYRK1A, the molecular "brake" that keeps beta cells quiescent.² Harmine alone produces a modest proliferation rate (~1.5–3% of cells dividing).³ Two combinations sharply amplify this: adding a TGF-beta–pathway inhibitor raises the dividing fraction to 5–8% (sometimes 15–18%),³ and adding a GLP-1 receptor drug (the class that includes semaglutide and exenatide) drives the largest effect.⁴

What the evidence actually shows

The strongest data come from a 2024 study in which human islets were transplanted into mice. Using a tissue-clearing microscopy method (iDISCO+) to count cells directly, harmine plus exendin-4 increased human beta-cell mass by roughly four- to sevenfold over three months, reversed diabetes in the mice, and did not change alpha-cell mass — an encouraging safety signal.⁴ A separate line of work showed that human alpha cells can be reprogrammed (via the factors PDX1 and MAFA) into glucose-responsive insulin-secreting cells that relieved diabetes in mice for at least six months — proof-of-concept for transdifferentiation.⁵

Crucially, all of this efficacy is preclinical — in dishes and in mice carrying human tissue, not in people with diabetes.¹⁴

Human data so far: safety only

The only completed human trial tested safety, not regeneration. In a Phase 1 single-ascending-dose study (NCT05526430), 25 healthy adults received pure pharmaceutical-grade oral harmine.⁶ It was generally tolerated, with no hallucinations or significant psychoactive effects below about 2.7 mg/kg; higher doses caused dose-dependent nausea and mild sensory effects.⁶ This study did not measure beta-cell mass and did not include people with diabetes.⁶

The unsolved problem: autoimmunity

Even a perfect regrowth drug faces a hard wall in T1D: the immune system that destroyed the original beta cells will attack new ones too. The researchers' patent disclosure states that regeneration will have to be paired with immunomodulators to last.⁴ No such pairing has yet been tested in humans. Off-target growth (DYRK1A acts in many tissues) and the safety of months-long dosing also remain open.¹⁶

Eligibility and durability

Because the strategy relies on multiplying existing cells, it is most plausible for people who still have residual beta cells — and durability is unproven in anyone.¹

What's coming

Next steps are next-generation, more beta-cell-selective DYRK1A inhibitors; trials that actually measure beta-cell function (e.g. C-peptide) in people; and — most importantly — combination studies that add immune protection so regrown cells can survive.¹⁴ This is one of the most elegant routes to a cure in concept and one of the least mature in practice. We rank it honestly as early-stage and will upgrade it the moment human efficacy or a regeneration-plus-immunotherapy trial reports.

As of our July 2026 review, a registry check confirms that is still where things stand: there is no harmine or DYRK1A-inhibitor trial in people with diabetes, and no human DYRK1A data appeared at the ADA 2026 scientific sessions.⁷⁸ Mount Sinai has said beta-cell-selective successors to harmine are moving through toxicology toward a first-in-human study, but none is registered and no date has been given — so we are not putting one on the calendar.

Meanwhile a different regeneration mechanism has gone further in people. Icovamenib, an oral menin inhibitor, produced a reported human C-peptide gain in 2026 — in five people, in an open-label trial with no completed placebo arm. That is a genuine signal, not a controlled result, and it says nothing about whether harmine works; but it does show the regeneration idea can be tested in people. Our icovamenib record covers it in detail.