Low-dose anti-thymocyte globulin (ATG)

What this evidence can establish

Rabbit anti-thymocyte globulin (ATG; Thymoglobulin) is an immune-depleting polyclonal antibody preparation. The current US label covers prevention and treatment of acute kidney-transplant rejection with other immunosuppression, not T1D. Changes in immune-cell populations do not mean ATG selectively removes only the cells attacking beta cells. The prevention and postdiagnosis studies below answer different clinical questions.⁷²

Randomized evidence after diagnosis

In TrialNet's recent-onset study, 89 participants were randomized to ATG (29), ATG plus GCSF (29) or placebo (31); eligibility was ages 12–45 and diagnosis within 100 days. One combination-group participant withdrew before treatment, leaving 88 in the safety analysis, and 87 completed the one-year primary measurement. Adjusted stimulated C-peptide at one year was 0.646 nmol/L with ATG versus 0.406 with placebo (one-sided P=0.0003). The combination comparison had P=0.031 and did not meet the prespecified 0.025 threshold. HbA1c was lower with ATG; insulin doses did not differ significantly. This measures retained insulin secretion after diagnosis, not freedom from insulin or delayed onset.¹

At two years, 82 of the original 89 participants completed the mixed-meal test. Adjusted C-peptide was 0.500 nmol/L (95% CI 0.412–0.594) with ATG versus 0.253 (0.177–0.334) with placebo. HbA1c remained lower, while insulin dose did not differ. This is follow-up of the same trial, not two independent two-year trials. The paper's abstract/Results and Figure 2 disagree on the exact C-peptide P value; the estimates and confidence intervals are reported here without resolving that discrepancy.²

The 2025 MELD-ATG trial randomized 117 participants aged 5–25 with recent-onset stage-3 T1D among placebo and four ATG dose groups (114 treated in the safety analysis). Both 2.5 mg/kg and 0.5 mg/kg showed better 12-month stimulated C-peptide than placebo. On the study's transformed ln(AUC + 1) scale, adjusted differences were 0.124 (95% CI 0.043–0.205) and 0.102 (0.021–0.183), respectively. The lower dose was identified through an adaptive dose-finding design; similar estimates do not demonstrate equivalence or noninferiority between doses. This review read the published abstract and the full accepted manuscript; typeset published tables were not separately retrieved.¹⁰

An earlier, separate 58-person trial of 6.5 mg/kg ATG did not improve its 12-month primary C-peptide outcome (P=0.591). Serum sickness affected all 38 treated participants and cytokine-release syndrome affected 37/38 in the detailed Results. The abstract's wording about those events differs slightly. This historical negative result matters, but comparisons across distinct trials do not prove that lowering a dose alone caused better efficacy.⁴

Harms are not just a tolerability footnote

In the low-dose TrialNet trial, serum sickness affected 21/29 ATG recipients, including 15/29 with grade-3 events (the report states none were grade 4); 17 of those 21 received steroids. Cytokine-release syndrome affected 14/29. No serious infection was reported in an ATG-containing arm, but the report also records a papillary thyroid tumor in one ATG recipient one month after the two-year assessment; this does not establish causation. MELD's abstract reports serum sickness in 27/33 at 2.5 mg/kg versus 11/34 at 0.5 mg/kg, and cytokine-release syndrome in 11/33 versus 8/34. Those lower observed rates do not establish an absence of serious or longer-term risk.¹¹⁰

The June 2026 US label warns of potentially fatal anaphylaxis or severe cytokine-release reactions, cytopenias, infection and malignancy, and advises against live vaccines soon after treatment. Its new thrombotic-microangiopathy warning concerns kidney-transplant recipients, particularly with calcineurin inhibitors. These warnings identify relevant hazards, but transplant-regimen event rates cannot be used as estimates for a low-dose T1D course.⁷

Stage 2: small reports, no established prevention benefit

STOP-T1D enrolled only two participants before termination for low enrollment and inability to meet its target. Results posted 7 May 2026 describe one ATG and one placebo participant, with 16 and 87 days respectively between randomization and diagnosis. The registry explicitly states that no outcome analyses were performed. It reports no deaths or serious adverse events in either one-person arm, but grade-4 lymphocyte reduction and grade-2 serum sickness in the ATG participant. These observations cannot establish comparative prevention efficacy or safety.⁸

A 2026 uncontrolled single-center report included 39 treated people (six stage 2 and 33 stage 3); only 22 had analyzable paired C-peptide measurements, including all six stage-2 participants. Its modified quantitative-response method used imputed C-peptide area under the curve and historical predictions, and was not validated for stage 2. Four of the six stage-2 participants met that exploratory response definition. This is not a randomized onset-delay result. Among the 22 analyzed participants, 18 experienced grade-2/3 serum sickness, 15 received steroids, and 11 had grade-1 cytokine-release syndrome.⁵

An earlier uncontrolled report from the same center described six children aged 5–14 with stage-2 T1D given off-label low-dose ATG: three remained diabetes-free after 1.5, 3 and 4 years of follow-up, while three progressed within one to two months of therapy. That report calls for prospective study rather than claiming efficacy.⁶ Because both reports come from the same center and group with no stated overlap accounting, their stage-2 participants must not be counted as independent evidence.

ASCEND/TN-40A (NCT07216391) is recruiting according to the registry update posted 9 September 2026. It is currently a randomized, open-label two-arm comparison of ATG with teplizumab in stage-2 T1D, ages 4 to under 35, with estimated enrollment 60. The primary endpoint is six-month change in the Diabetes Prevention Trial Risk Score (DPTRS), a risk measure rather than observed delay of diagnosis; clinical progression is a secondary outcome. Estimated primary completion is November 2029 and no results are posted. Registry recruitment status is not confirmation that any particular site can enroll a person today.⁹

Access, funding and score limits

The current US label does not include T1D. The 2026 US cohort describes insurer-authorized or self-funded off-label use: reported average out-of-pocket costs were under US$2,500, and its estimated self-pay cost was US$10,000–15,000 depending on dose. Those are one center's historical estimates, not current price quotes or evidence of affordable, funded global access. The study was publicly/charitably supported and disclosed relevant commercial relationships. TrialNet's randomized study had public/foundation funding and manufacturer-supplied drug; MELD reports European Innovative Medicines Initiative funding. Score limitation: the retained prevention-category numbers have not been calibrated separately for prediagnosis delay and postdiagnosis C-peptide preservation. The access number is not a verified price or coverage assessment.⁷⁵¹¹⁰