Low-dose anti-thymocyte globulin (ATG)

Screening: finding T1D before it strikes

Like every prevention therapy, low-dose ATG only helps people who are found in time — and that depends on screening. Type 1 diabetes develops along a defined three-stage path. Stage 1 is two or more islet autoantibodies (against insulin, GAD, IA-2, or ZnT8) with normal blood glucose and no symptoms; stage 2 adds abnormal glucose (dysglycemia) but still no symptoms; stage 3 is clinical disease, when insulin becomes necessary.¹ These autoantibodies are powerful predictors: in genetically at-risk children with two or more, symptomatic diabetes develops in roughly 44% by 5 years, 70% by 10 years, and 84% by 15 years, with lifetime risk approaching 100% — so the staging model is reliable enough to build prevention trials on.¹

Screening is often offered to relatives of someone with T1D, but about 90% of new cases have no family history — so finding most at-risk people requires general-population autoantibody (and sometimes genetic-risk) testing.² Early detection matters even without treatment: people identified through screening are monitored, so when they reach stage 3 they are far less likely to arrive in diabetic ketoacidosis (DKA), a dangerous and sometimes fatal emergency.² Capillary and home-collected blood-spot methods are lowering the cost and burden that long blocked wide screening.²

Therapy: an old transplant drug, repurposed and shrunk

Anti-thymocyte globulin is a polyclonal antibody (rabbit-derived Thymoglobulin, from Sanofi/Genzyme) that has been used for decades to deplete T cells in organ-transplant patients. The repurposing idea is simple: T1D is driven by T cells destroying insulin-producing cells, so depleting those T cells should slow the attack.³ The breakthrough was dose. A high-dose course (6.5 mg/kg) tested in new-onset T1D failed to preserve insulin-producing function and caused serum sickness and cytokine-release syndrome in nearly every patient.⁴ Counter-intuitively, a much lower dose works better — it shifts the balance toward protective regulatory T cells (raising the Treg-to-effector ratio) rather than just blunt depletion.³

The pivotal evidence is the TrialNet phase 2b trial (NCT02215200): 89 people aged 12–45, within 100 days of diagnosis, randomized to low-dose ATG (2.5 mg/kg), ATG plus GCSF, or placebo. At one year, stimulated C-peptide (a direct readout of surviving insulin-producing cells) was 0.646 nmol/L with ATG versus 0.406 with placebo (p=0.0003), and HbA1c was significantly lower.⁵ At two years the benefit held (C-peptide p=0.00005; HbA1c still lower), with the decline in C-peptide roughly halved versus placebo — durable for an immunotherapy and from a single course.³ Adding GCSF gave no extra benefit.

The decisive follow-up is the 2025 MELD-ATG trial (NCT04509791), an INNODIA dose-ranging study of 117 people aged 5–25 with recent-onset (stage 3) T1D across eight European countries.⁶ It found that an even lower 0.5 mg/kg dose preserved C-peptide about as well as 2.5 mg/kg — and far more safely: serum sickness fell from 82% to 32%, and cytokine-release from 33% to 24%.⁶ That makes ATG a rare thing in T1D: an affordable, off-patent, globally stocked disease-modifying drug.⁶

Safety: the main effects are serum sickness and cytokine-release symptoms in the days after infusion — uncomfortable but predictable and short-lived, with no deaths and no long-term safety signals in these trials.³ The lower dose meaningfully shrinks this burden.

Access: ATG is widely available and cheap, but it is not approved for T1D anywhere — use is investigational or off-label, given as a brief IV infusion over two days, and (like all prevention therapy) depends on screening to find candidates.⁵

The prevention question — and why it stalled

Every trial above treats people who are already diagnosed (stage 3). The stated next step, voiced by the investigators themselves, is to move earlier — into stage 1/2 to delay or prevent clinical onset, the same territory where teplizumab is already approved.³ With the safe minimum dose now defined by MELD-ATG, a low-cost repurposed agent looks well positioned for exactly that.⁶

It has not gone to plan. The dedicated stage-2 prevention trial — STOP-T1D (TN28, NCT04291703), run by NIDDK's TrialNet — opened 22 sites across the US and Australia, aimed to randomize people aged 6–34 who had stage-2 T1D and a roughly 50% two-year risk of progressing, and ran from November 2023 to December 2024. It enrolled two participants, and was terminated about 13 months after the first enrolment — the registry record gives the reason as "low enrollment and lack of feasibility to meet target enrollment."⁷

Two people across 22 sites is the whole story. This is a recruitment failure, not a scientific one — nothing in the result says ATG does not work in stage 2, because the trial never got the chance to ask. But it is an honest measure of how hard prevention trials are to run. Stage 2 is asymptomatic: candidates have to first be found by autoantibody screening (which most countries do not do at scale), then persuaded to take an infusion of an immune-depleting drug — with serum sickness a realistic side effect — for a disease they do not yet have and might, in their own mind, never get. It does not help that teplizumab is already approved for stage 2, which makes being randomized to an unapproved alternative a harder thing to say yes to.

What's coming

The stage-2 question has been restructured, not abandoned. It now sits inside TrialNet's adaptive platform trial TN-40A (NCT07216391), which pits low-dose ATG head-to-head against teplizumab in stage-2 T1D (ages 4–34), with a start date of 30 July 2026.⁸ A platform trial shares infrastructure and a control arm across multiple candidate drugs, which is precisely the design you reach for when single-drug recruitment has proved impossible — and running ATG against the approved therapy, rather than against placebo, removes the main reason a family would decline.

That trial is the real test. If a cheap, off-patent, globally stocked drug can match an expensive new biologic at delaying onset, prevention could reach far more people than teplizumab alone ever will. Until TN-40A reports, ATG's prevention benefit remains unproven — its evidence base is entirely in people who already have stage 3 disease.

References

1. Insel RA, et al. Staging presymptomatic type 1 diabetes: a scientific statement of JDRF, the Endocrine Society, and the American Diabetes Association. Diabetes Care (2015). https://doi.org/10.2337/dc15-1419 ↩ ↩²

2. Sims EK, et al. Screening for Type 1 Diabetes in the General Population: A Status Report and Perspective. Diabetes (2022). https://doi.org/10.2337/dbi20-0054 ↩ ↩² ↩³

3. Haller MJ, et al. Low-Dose Anti-Thymocyte Globulin Preserves C-Peptide, Reduces HbA1c, and Increases Regulatory to Conventional T-Cell Ratios in New-Onset Type 1 Diabetes: Two-Year Clinical Trial Data. Diabetes (2019). https://doi.org/10.2337/db19-0057 ↩ ↩² ↩³ ↩⁴ ↩⁵

4. Gitelman SE, et al. Antithymocyte globulin treatment for patients with recent-onset type 1 diabetes: 12-month results of a randomised, placebo-controlled, phase 2 trial. Lancet Diabetes Endocrinol (2013). https://doi.org/10.1016/S2213-8587%2813%2970065-2 ↩

5. Haller MJ, et al. Low-Dose Anti-Thymocyte Globulin (ATG) Preserves β-Cell Function and Improves HbA1c in New-Onset Type 1 Diabetes. Diabetes Care (2018). https://doi.org/10.2337/dc18-0494 ↩ ↩²

6. Mathieu C, et al. Minimum effective low dose of antithymocyte globulin in people aged 5–25 years with recent-onset stage 3 type 1 diabetes (MELD-ATG): a phase 2, multicentre, double-blind, randomised, placebo-controlled, adaptive dose-ranging trial. Lancet (2025). https://doi.org/10.1016/S0140-6736%2825%2901674-5 ↩ ↩² ↩³ ↩⁴

7. STOP-T1D Low-Dose ATG (TN28) — TERMINATED. NCT04291703, NIDDK/TrialNet. ClinicalTrials.gov (accessed 2026-07-14). https://clinicaltrials.gov/study/NCT04291703 ↩

8. TrialNet platform TN-40A: teplizumab vs low-dose ATG in stage-2 type 1 diabetes. NCT07216391, NIDDK/TrialNet. ClinicalTrials.gov (accessed 2026-07-14). https://clinicaltrials.gov/study/NCT07216391 ↩