Two small randomized trials have now tested a GLP-1 drug alongside modern insulin therapy in type 1 diabetes, and both found something worth knowing. Neither makes it an approved treatment, and the reason to read them carefully is that the changes they describe involve taking insulin away from people who cannot live without it.
What happened to the insulin
A post hoc analysis of the ADJUST-T1D trial followed adults with type 1 diabetes and obesity who were using automated insulin delivery systems, and who took semaglutide 1 mg weekly or placebo for 26 weeks [1]. Total daily insulin fell 22.6%, with an interval running from 28.3 to 17.0 — a change of the kind this site sizes carefully in what each dose step took away.
The reduction was not spread evenly across the two kinds of insulin a person takes. Bolus doses, the ones covering meals, fell 30.5%, while basal doses covering the background requirement fell 15.6%. The share of the total that was basal therefore rose from 0.56 to 0.62, and insulin per kilogram per day fell from 0.72 to 0.60. Daily carbohydrate intake fell from 137 g to 107 g over the same period.
What happened to the glucose
A separate double-blind crossover trial gave weekly semaglutide or placebo to people using automated insulin delivery, randomizing 28 participants of whom 24 completed [2]. Time in the target glucose range rose by a mean 4.8 percentage points against placebo.
| Outcome | Result | Significance |
|---|---|---|
| Time in target range | +4.8 points (SD 7.6) | P = 0.006 |
| Time below 3.9 mmol/L | no increase | P = 0.19 |
| Time below 3.0 mmol/L | no increase | P = 0.65 |
Improving time in range without buying it back in low readings is the combination that matters in type 1 diabetes, because a drug that flattens glucose by pushing people toward hypoglycemia has not helped them at all.
The safety signal a type 1 reader needs
How far this goes
Not far, and the numbers are the reason. Twenty-eight people were randomized in one trial, and the other is a post hoc analysis of a larger but still modest study. Neither is the scale at which a safety signal separates from chance, or at which subgroups become visible.
Both trials also enrolled people already using automated insulin delivery, which is a system that adjusts doses continuously and catches drift that a fixed regimen would not. Whether the same reductions would be safe on injections and finger-sticks is untested, and it is the setting most people with type 1 diabetes are actually in.
What happened when it stopped
The ADJUST-T1D team kept continuous glucose monitors on after the 26-week trial ended and followed people for a 12-week extension [3]. Sixteen had come off semaglutide, and 22 had been on placebo and started no GLP-1 drug afterwards, which is a small group on either side.
Time in range fell by a median 3.6 percentage points in the people who stopped, against a 1.5-point rise in the placebo group. Glucose also swung more: the standard deviation rose 6.0 mg/dL against 1.2. Time above range crept up but lost significance after adjustment, and hypoglycemia measures did not differ between the groups. The authors read it as benefits that may not fully survive withdrawal, which is the story told at more length in what happened when type 1 patients stopped. Nothing here is a reason to change an insulin dose, which is a decision that belongs to the clinician managing it.