A groundbreaking study indicates that type 1 diabetes may not be a single disease but two genetically and biologically distinct conditions, potentially transforming diagnosis and treatment approaches.
Type 1 diabetes may need to be understood not as a single disorder but as at least two genetically distinct forms, with one appearing to be driven more by mast-cell biology and the other more by T-cell activity, according to a large international study published in Diabetologia on 31 August 2026.
The paper, led by researchers in the US and UK, is described as the first genome-wide association study to split type 1 diabetes by the two major high-risk HLA backgrounds known as DR3 and DR4. HLA genes help control immune responses, and these two patterns have long been linked to the disease. By separating them, the researchers argue, scientists may be seeing not just different routes into the same diagnosis but genetically and biologically distinct forms of illness.
To test that idea, the team analysed 9,091 people with type 1 diabetes and 14,157 controls drawn from multiple cohorts. The main analysis was restricted to people of European ancestry, where DR3 and DR4 are most common. After excluding those who carried both DR3 and DR4, because they could not be cleanly assigned to one group, the researchers compared 4,057 DR3 cases with 8,043 controls and 5,034 DR4 cases with 6,114 controls.
The distinction matters because DR3 and DR4 were already known to track with different early immune signals. In the paper, the authors say DR3 is associated with a 3.5-fold higher rate of glutamic acid decarboxylase autoantibodies appearing first, a pattern linked to later onset. DR4, by contrast, is associated with a 2.4-fold higher rate of insulin autoantibodies appearing first, which is more often tied to earlier disease. As the researchers put it, “Type 1 diabetes is a complex disease characterised by autoimmune destruction of beta cells, but the underlying aetiology is not well understood.”
What is new here is the scale of the genetic split. The authors reported only a moderate genetic correlation between DR3-linked and DR4-linked disease, lower than the similarity seen when type 1 diabetes is divided by age at diagnosis or by sex. In the paper, they said that level of separation was comparable to the genetic relationship seen between paired disorders such as Crohn’s disease and ulcerative colitis, and even schizophrenia and bipolar disorder. That does not prove these are already two formally separate diagnoses, but it does suggest the usual umbrella label may be hiding important differences.
The gene-level findings also point in that direction. Eleven loci reached genome-wide significance in one or both analyses. Seven were shared across both groups, including well-known type 1 diabetes genes such as PTPN22, IL2RA and INS. But some signals were more specific: CTLA4 and HORMAD2 reached that threshold only in DR3-linked disease, while CLEC16A and IL2 did so only in DR4-linked disease. Of those, IL2 stood out because its effect was significantly stronger in the DR4 group, strengthening the idea that the immune circuitry is not identical across the two forms.
When the researchers looked beyond individual genes to regulatory DNA regions that help switch genes on and off, the divergence became more biologically concrete. DR4-linked disease showed stronger enrichment in T-cell regulatory elements and T-cell-related pathways, which fits the conventional picture of immune cells attacking insulin-producing beta cells in the pancreas. DR3-linked disease, however, was specifically enriched in mast-cell regulatory elements and secretion-related pathways. Mast cells are better known for their role in allergy and inflammation, so that signal opens a less familiar line of inquiry for type 1 diabetes research.
That split could eventually matter in the clinic as much as in the laboratory. The authors note that people with DR4 have already appeared more responsive to preventive treatments aimed at T-cell activity, including methyldopa and teplizumab. If confirmed, that would mean a single prevention strategy may not work equally well for everyone now grouped under type 1 diabetes. Several reports on the study highlighted the same practical point: identifying a more precise subtype could help doctors tailor treatment and help researchers design sharper trials.
The researchers were careful not to overstate the case. DR3 and DR4 were used as proxies for the first autoantibody to appear, and the link is only partial rather than exact. The study was also limited to European-ancestry samples, so the findings may not transfer neatly to populations in which other HLA risk haplotypes are more common. The paper further notes that people carrying both DR3 and DR4, who are at especially high risk, were left out of the main head-to-head split. Even so, the authors argue that future genetic studies, risk-prediction models and clinical trials should start taking DR3 and DR4 status into account. If that happens, type 1 diabetes may come to be classified less as one uniform autoimmune disease and more as a set of overlapping pathways that end in the same diagnosis.
Disclaimer: This content is for informational purposes only and is not intended to be a substitute for professional medical judgment, advice, diagnosis, or treatment.
