Semaglutide is not approved for type 1 diabetes. Everyone in this study was in a trial, and the question it answers is narrow: what happens to blood sugar control in the twelve weeks after the drug is taken away. On the weight side that question already has a well-established answer, since regain after discontinuation is the rule rather than the exception. This is the glucose version, in a different population.
The setup
Participants were adults with type 1 diabetes and obesity using an automated insulin delivery system — a pump that adjusts insulin itself, minute by minute, from a continuous glucose sensor. [1] The extension phase followed them after semaglutide or placebo stopped.
Time in range means the share of the day spent between 70 and 180 mg/dL. It is the number that matters most to somebody wearing a sensor.
What changed
Time in range fell by a median 3.6 points in the group that had been on semaglutide. Placebo went up 1.5 points. Adjusted p = 0.044.
Variability rose as well. The standard deviation of glucose went up 6.0 mg/dL against 1.2. The coefficient of variation went up 3.1% against 1.1%. Both at adjusted p = 0.044.
What did not change
Nothing in the low-blood-sugar measures moved. That is the finding worth holding onto.
The standing worry about combining a GLP-1 with insulin is that the two together drive glucose too low. Stopping the GLP-1 did not shift those metrics in either direction here, which is a null result rather than a reassurance — and a null in a small extension study carries the weight its sample allows.
How big is 3.6 points
A little under an hour a day. That is the honest translation of three and a half percentage points of a twenty-four-hour clock.
Whether an hour a day out of range matters is a clinical judgment, not a statistical one, and it is the same distinction as an effect that clears significance and sits below perception. Here the direction is clear and the size is modest.
What it says about stopping generally
That the benefit was not retained. Twelve weeks off the drug and the measurements drifted back toward where they started.
That matches the weight side exactly: different population, different endpoint, same shape. A placebo arm is what makes the comparison mean anything, which is not true of every study that reports a before and an after.