TN-07: Oral insulin for prevention in autoantibody-positive relatives

What was tested, and why it matters

Type 1 diabetes develops when the immune system attacks the insulin-making cells in the pancreas. Years before any symptoms, that attack shows up as islet autoantibodies in the blood — so relatives of people with type 1 diabetes can be screened and flagged as "at risk."¹ TN-07 asked a hopeful question: if you feed the immune system a small daily dose of insulin by mouth, can you gently retrain it to tolerate insulin instead of attacking it — and so delay or prevent the disease?¹ Oral insulin is digested, not absorbed as a hormone, so the goal was immune "tolerance," not blood-sugar lowering.¹

Who it was for

The trial enrolled relatives of people with type 1 diabetes who already had at least two islet autoantibodies (including insulin autoantibodies) but whose glucose tests were still normal — people in the silent, pre-symptomatic stages of the disease.¹ Participants ranged from young children to adults (ages 3–45; median enrollment age about 8 years).²

How it was designed

TN-07 was a Phase 3, randomized, triple-masked, placebo-controlled trial run across nine countries (Canada, the US, Australia, New Zealand, the UK, Italy, Sweden, Finland, and Germany).¹² In total 560 people were randomized to either 7.5 mg of oral insulin daily or a matching placebo.¹ They were sorted into groups based on their antibody profile and insulin-response strength; the main analysis group had 389 participants, followed for a median of 2.7 years.¹

The key results

In the main study group, oral insulin did not delay type 1 diabetes: diabetes developed in 28.5% on oral insulin versus 33% on placebo (hazard ratio 0.87; P=.21 — not significant).¹ Across the whole cohort the result was likewise not significant.¹

But one pre-specified subgroup stood out — relatives whose insulin response was already weakening (secondary stratum 1, n=55). There, the publication reports 13 diabetes diagnoses on oral insulin versus 19 on placebo, and time to diabetes was significantly longer (HR 0.45; P=.006).¹ No significant study-related adverse events were reported. Although the stratum was prespecified, secondary analyses were not adjusted for multiple comparisons and must be interpreted as exploratory.¹

What it means and what's next

The headline was negative — oral insulin as used here does not prevent type 1 diabetes — but the subgroup signal kept the antigen-tolerance idea alive.¹ Later analyses of TN-07 data sharpened the picture: participants with high IA-2 autoantibody levels (HR 0.62; P=.012) and carriers of the HLA DR4-DQ8 genetic type (HR 0.59; P=.027) appeared to have their diabetes delayed, suggesting a specific "endotype" that may respond to oral insulin.³ The lesson shaping today's prevention research: the right therapy may need to be matched to the right person, identified early by their antibodies and genetics.³

Source-count limitation: The primary publication’s abstract reports subgroup percentages that do not consistently reconcile with its Table 1 allocation counts. We report the published event counts and hazard ratio rather than silently recalculating a disputed denominator. The overall primary result remains negative. Primary JAMA report.