Semaglutide in Type 1 Diabetes: Why AID Changes the Math

Semaglutide produced meaningful weight loss and better continuous-glucose-monitoring outcomes in one narrow group of people with type 1 diabetes: adults with obesity who already used automated insulin delivery (AID). In the 26-week ADJUST-T1D trial, 36% of participants assigned semaglutide met a demanding combined target, compared with 0% assigned placebo. That is a real result. It is also not a license to treat type 1 diabetes broadly with GLP-1 drugs.

The distinction matters because earlier randomized evidence found that GLP-1 receptor agonists can lower average glucose in type 1 diabetes while increasing time spent low. AID may be doing much of the safety work in the newer trial by automatically adjusting or suspending insulin as glucose falls. Semaglutide in type 1 diabetes is therefore best understood as an off-label, specialist-managed option under study for people with both obesity and closed-loop insulin delivery—not as a replacement for insulin or a general-purpose glucose fix.

A small randomized trial with a meaningful composite result

The 2025 ADJUST-T1D trial randomly assigned 72 adults with type 1 diabetes, a BMI of at least 30, and AID use to weekly semaglutide, titrated to 1 mg, or placebo. The double-blind trial lasted 26 weeks, with 36 participants in each group.

Its primary endpoint required participants to achieve all three of the following: more than 70% time in range, less than 4% time below range, and at least 5% weight loss. Semaglutide met that endpoint in 36% of participants versus none on placebo, an absolute between-group difference of 36 percentage points (95% CI, 20.6 to 52.2; P<0.001).

The individual outcomes moved in the same direction. Compared with placebo, semaglutide reduced HbA1c by 0.3 percentage points (95% CI, -0.6 to -0.05), increased time in range by 8.8 percentage points (95% CI, 3.9 to 13.7), and reduced body weight by 8.8 kg (95% CI, -10.6 to -7.0). The weight result carries much of the clinical interest. Obesity can add insulin resistance and cardiovascular risk to the existing burden of type 1 diabetes, and treatment options for obesity in this population remain limited.

Severe hypoglycemia occurred twice in each treatment group, and the trial reported no diabetic ketoacidosis (DKA). Those findings are reassuring within this small, carefully selected study. They cannot establish that semaglutide carries no DKA risk in routine care.

Why average glucose can mislead

HbA1c can fall because glucose values shift downward overall, including into hypoglycemic territory. That is why the ADJUST-T1D composite included both time in range and a ceiling on time below range. It asked whether participants could improve glucose control without buying that improvement with more lows.

The broader evidence explains why that design was sensible. A 2024 meta-analysis of six randomized trials, involving 378 adults with type 1 diabetes, found that adjunctive GLP-1 receptor agonists reduced HbA1c by 0.21 percentage points versus control (95% CI, -0.36 to -0.06; P=0.007) and reduced time above range by 1.83 percentage points (95% CI, -2.51 to -1.15; P<0.001).

But time in range did not improve significantly: the mean difference was -0.22 percentage points (95% CI, -2.39 to 1.95; P=0.84). Time below range increased by 1.13 percentage points (95% CI, 0.50 to 1.76; P<0.001). Measures of glucose variability, including mean amplitude of glycemic excursions, coefficient of variation, and standard deviation, also showed no significant improvement.

These results do not contradict the ADJUST-T1D trial so much as define its boundary. The meta-analysis pooled several agents and heterogeneous type 1 diabetes populations, most without AID. ADJUST-T1D studied one drug in people using closed-loop insulin delivery. AID can respond to a falling glucose trajectory in real time; multiple daily injections and standalone CGM cannot autonomously reduce insulin. The favorable hypoglycemia profile in ADJUST-T1D may therefore depend on the device as much as on the medication.

Semaglutide did not make insulin optional. It added appetite and post-meal effects to an insulin system that was already automating part of the safety problem.

What semaglutide can—and cannot—do in type 1 diabetes

In type 1 diabetes, semaglutide cannot rely on the usual goal of stimulating meaningful endogenous insulin secretion from beta cells. Its plausible benefits are extra-pancreatic: suppressing post-meal glucagon, slowing gastric emptying, and reducing appetite through central GLP-1 signaling. Less food intake can reduce body weight and insulin requirements; slower nutrient absorption and lower glucagon can alter post-meal glucose excursions.

Those mechanisms are plausible, but they also create practical hazards. Appetite suppression may lead someone to reduce meal insulin aggressively. If insulin falls too far, ketone production can increase even when glucose does not look dramatically high. Euglycemic DKA is a mechanistic concern, not a complication ruled out by 18 patient-years of semaglutide exposure in ADJUST-T1D. Anyone prescribed a GLP-1 drug off label for type 1 diabetes needs a clinician-directed plan for insulin adjustment, ketone monitoring, carbohydrate intake, and symptoms that warrant urgent evaluation.

What real-world and pediatric data add

A 2025 real-world study in overweight and obese people with type 1 diabetes using AID reported glycemic and weight signals consistent with ADJUST-T1D. Because it was observational, it cannot separate drug effects from selection, changes in clinical care, or other differences between people who did and did not receive treatment. Its value is external context: this patient profile exists outside a trial, and clinicians are already prescribing these drugs off label.

Evidence in younger people is thinner. A 2026 retrospective cohort study of youth with type 1 diabetes associated GLP-1 receptor agonist use with lower BMI and lower total daily insulin dose. The uncontrolled design and small safety evidence leave major questions unanswered, especially about DKA risk in adolescents.

Cardiovascular side benefits remain unproven in type 1 diabetes

GLP-1 drugs may offer effects beyond weight and glucose, but type 2 diabetes evidence should not quietly become type 1 diabetes evidence. A 2025 meta-analysis in type 2 diabetes found reductions in inflammatory markers including CRP, IL-6, and TNF-alpha. A separate 2024 semaglutide meta-analysis reported 42% lower incident atrial fibrillation versus placebo (RR 0.58, 95% CI, 0.40 to 0.85) across 10 randomized trials in diabetic and high-risk populations. Neither finding establishes an anti-inflammatory or anti-arrhythmic benefit in type 1 diabetes.

The practical verdict

For an adult with type 1 diabetes, BMI at least 30, and active AID use, ADJUST-T1D provides a reasonable proof of concept: weekly semaglutide improved a combined glucose-and-weight endpoint over 26 weeks, with an average 8.8 kg greater weight loss than placebo. The trial was small, short, and funded by Breakthrough T1D, a patient advocacy and research organization. Its findings need a larger, longer confirmation trial that can assess uncommon but serious harms.

For people with type 1 diabetes who do not have obesity or do not use AID, the stronger applicable warning comes from the randomized meta-analysis: modest HbA1c lowering without a significant time-in-range gain, alongside more time below range. Semaglutide may eventually fill an important treatment gap for obesity in selected people with type 1 diabetes. Today, it remains an off-label intervention whose best evidence depends on both patient selection and automated insulin delivery.