GAD-alum antigen-specific immunotherapy (Diamyd)

Screening: who is found, and why it matters

Type 1 diabetes (T1D) has a long silent phase before symptoms. The immune system can be caught attacking the insulin-producing beta cells years early by testing for islet autoantibodies — antibodies against insulin, GAD65, IA-2, and ZnT8. The field uses a three-stage model: stage 1 is two or more autoantibodies with normal blood sugar; stage 2 adds abnormal (dysglycemic) blood sugar but still no symptoms; stage 3 is clinical diabetes needing insulin.⁵ Two or more autoantibodies is a powerful signal: in pooled cohorts, about 70% of such children progressed to diabetes within 10 years.⁶ Finding people early lets families act before a crisis — and early detection opens a window for disease-modifying therapy.⁵

For GAD-alum specifically, screening also asked a genetic question. Earlier evidence suggested the therapy worked only in people carrying the HLA DR3-DQ2 tissue type — roughly 40% of the T1D population in Europe and the US — so the phase-3 trial screened candidates for both GAD65 antibodies and this genotype, and enrolled only DR3-DQ2 carriers.⁷⁸ That bet is the one the phase 3 tested, and lost.

Therapy: what GAD-alum did, and why the phase 3 failed

GAD-alum (retogatein: recombinant GAD65 protein bound to an aluminium adjuvant) is an antigen-specific immunotherapy: instead of suppressing the whole immune system, it presents one of the proteins the immune system mistakenly targets, aiming to teach tolerance and spare beta cells.⁹ It was given by injection directly into a lymph node (intralymphatic), three doses a month apart, alongside oral vitamin D.⁹

The earlier picture was mixed-then-focused. An early subcutaneous trial in recent-onset patients missed its main C-peptide endpoint but hinted at preserved insulin secretion.¹⁰ The phase IIb DIAGNODE-2 trial (NCT03345004; 109 patients aged 12-24) again missed its primary endpoint in the full group, but in the pre-specified HLA DR3-DQ2 subgroup it preserved stimulated C-peptide markedly better than placebo (treatment effect ratio 1.56, p=0.008 at 15 months).⁹ A pooled re-analysis of three randomized trials found a significant, dose-dependent benefit confined to DR3-DQ2 carriers.⁸

That signal did not replicate. The phase 3, DIAGNODE-3 (NCT05018585), was built entirely around it: it enrolled only HLA DR3-DQ2 carriers — genotyping was a screening gate, so the whole 321-person trial population was the group the therapy was supposed to help.⁷ On 27 March 2026 Diamyd reported that the pre-specified interim analysis (174 participants, baseline to month 15) missed the primary C-peptide endpoint and failed the pre-specified criteria for continuing.¹ On 10 April 2026, after independent external statistical validation, the company discontinued the trial for futility: no clinically meaningful effect on C-peptide in the trial population or in any further pre-specified subgroup, and no difference in HbA1c or time-in-range.² The precision-medicine thesis was tested on its own chosen terms, and it failed.

The one thing that held up is safety: no safety concerns were identified at the interim review or at discontinuation, and across all the trials the therapy caused only minor, transient injection-site reactions.⁹² The failure was efficacy, not harm.

Access and what's coming

GAD-alum is not approved anywhere, and is no longer in active clinical development. The accelerated-approval route the programme was aiming for — the FDA had granted Fast Track and Orphan Drug designations — depended on a positive interim readout from DIAGNODE-3.¹¹ That readout was negative, so the route is gone.¹²

Diamyd started a formal strategic review after the discontinuation, and on 11 May 2026 said it does not intend to allocate further internal resources to retogatein's clinical development, and is preparing the data for possible out-licensing, partnership or transfer.³

If you are living with T1D and were watching this one: it is a genuine loss, and it is worth being clear about what it means. Antigen-specific immunotherapy — teaching tolerance to one protein rather than blunting the whole immune system — remains an attractive idea, and this result does not kill the idea. But it is a hard reminder that a striking subgroup finding in a phase 2 trial is a hypothesis, not a result, and that the honest test is a properly powered trial in exactly that group. This one got that test.