Stopping Semaglutide Changed Glucose Control in Type 1 Diabetes. The Study Was Small.

A new follow-up to the ADJUST-T1D trial offers a rare look at what happened after adults with type 1 diabetes and obesity stopped semaglutide. Over the next 12 weeks, their continuous glucose monitor data moved in the wrong direction: less time in range and more day-to-day glucose variability.

That result sounds like a clean warning about withdrawal. It is not. Only 16 people were in the group that stopped semaglutide, and the extension was not a fresh randomized withdrawal trial. Still, it addresses a practical question that most drug trials leave unanswered. Benefits are measured while treatment continues. The period after treatment ends often gets little attention.

For people with type 1 diabetes, that gap matters. Semaglutide can affect appetite, body weight, glucose patterns, and insulin requirements. A person using an automated insulin delivery system may have technology adjusting insulin in the background, but the system still depends on changing physiology, meal patterns, settings, and user decisions. Stopping a drug that altered several of those variables at once is not a neutral event.

What the extension study found

The analysis followed participants who had completed ADJUST-T1D, a 26-week randomized trial of once-weekly semaglutide in adults with type 1 diabetes, obesity, and automated insulin delivery. During the 12-week extension, researchers collected continuous glucose monitoring data from two groups: 16 participants who had received semaglutide and then stopped it, and 22 participants originally assigned to placebo who did not use a GLP-1 receptor agonist during follow-up.

The main signal was a change in time in range, defined as glucose between 70 and 180 mg/dL. The former semaglutide group had a median decline of 3.6 percentage points. The former placebo group improved by 1.5 points. The adjusted p value was 0.044.

Glucose also became more variable after semaglutide withdrawal. Standard deviation rose by 6.0 mg/dL in the withdrawal group versus 1.2 mg/dL in the placebo group. Coefficient of variation increased by 3.1% versus 1.1%. Both comparisons had adjusted p values of 0.044, and an analysis that accounted for week 26 values produced consistent findings.

Not every measure crossed the statistical threshold. Time above 180 mg/dL increased numerically after withdrawal, but the comparison did not remain significant after adjustment. The adjusted p value was 0.064. Measures related to hypoglycemia did not differ significantly between groups.

That pattern is more informative than a simple claim that glucose “rebounded.” The extension detected a modest deterioration in range and stability. It did not show a statistically significant increase in hypoglycemia, and it did not establish how much of the change came from weight, appetite, insulin dosing, meal boluses, or a direct pharmacologic effect.

The parent trial explains why stopping could matter

ADJUST-T1D enrolled 72 adults with type 1 diabetes, a body mass index of at least 30, and an automated insulin delivery system. Participants were assigned in a 1:1 ratio to semaglutide, titrated up to 1 mg weekly, or placebo for 26 weeks.

The primary endpoint required participants to meet three goals at the same time: more than 70% time in range, less than 4% time below 70 mg/dL, and at least 5% weight loss. Thirty-six percent of the semaglutide group reached that composite endpoint, compared with none of the placebo group.

The individual outcomes help put that result in perspective. Compared with placebo, semaglutide improved HbA1c by 0.3 percentage points, increased time in range by 8.8 percentage points, and reduced body weight by 8.8 kilograms at week 26. Two severe hypoglycemia events occurred in each group, and the investigators reported no diabetic ketoacidosis.

A later post hoc analysis showed that insulin use changed quickly. Total daily insulin dose fell by 22.6% over 26 weeks with semaglutide. Bolus insulin decreased by 30.5%, while basal insulin fell by 15.6%. At week 4, the investigators attributed most of the dose reduction to a direct drug effect rather than weight loss. By week 26, the estimated contributions of drug effect and weight loss were roughly even.

Those insulin findings make the withdrawal study clinically interesting. If insulin needs fall after semaglutide starts, they may not stay at the same level after it stops. The extension paper did not provide a complete rule for reversing dose changes, and it should not be used as one. It does show why a stop date deserves active planning rather than a passive “see what happens” approach.

What automated insulin delivery can and cannot do

Automated insulin delivery systems combine continuous glucose monitoring with an algorithm that adjusts insulin delivery. They can respond to changing glucose, but they do not remove the need for clinical judgment. Systems differ, many still rely on user-entered meal information, and their performance is shaped by settings such as insulin-to-carbohydrate ratios and correction factors.

Semaglutide adds another moving part. During treatment, a person may eat less, lose weight, need less mealtime insulin, or experience delayed gastric emptying. After discontinuation, appetite and meal size may change again. Weight and insulin sensitivity may also shift over a longer period. An algorithm can react to glucose data, but it cannot know in advance why the inputs changed or whether current settings remain appropriate.

The new study supports closer CGM review during that transition. It does not prove that everyone will lose glycemic control, nor does it identify one universal monitoring schedule. The participants were adults with obesity who already used automated insulin delivery. Someone using injections, a conventional pump, or a different GLP-1 drug may have a different experience.

The limitations are substantial

The strongest fact in this paper is also its biggest constraint: only 38 people contributed to the extension analysis. Sixteen had stopped semaglutide. With samples that small, a few participants can influence the median, and uncommon safety outcomes are nearly impossible to assess.

The comparison also changed after the randomized phase ended. Participants were no longer being randomly assigned to withdrawal or continuation. The researchers compared people who stopped semaglutide with former placebo participants who remained off GLP-1 therapy. That is useful, but it is weaker than a blinded trial that rerandomizes responders to continue treatment or stop.

Follow-up lasted 12 weeks. The paper can describe short-term CGM changes, but it cannot tell us whether the decline stabilizes, worsens, or reverses later. The abstract does not report long-term weight regain, HbA1c, insulin dose, or appetite outcomes for this extension. It also cannot separate the loss of semaglutide's direct effects from behavioral and physiologic changes that follow discontinuation.

Finally, semaglutide is not established as routine glucose-lowering therapy for type 1 diabetes. The parent trial was conducted in a selected population with obesity and automated insulin delivery. Its findings should not be stretched into a general recommendation for all adults with type 1 diabetes.

The Oria take

The useful lesson is not that semaglutide must never be stopped. Cost, tolerability, access, pregnancy planning, lack of benefit, or a clinician's safety judgment can all change a treatment plan. The lesson is that discontinuation is part of treatment, and it needs the same attention as initiation.

For a person with type 1 diabetes, stopping semaglutide may coincide with rising insulin needs and less stable glucose. The ADJUST-T1D extension gives that concern a measurable signal, but not a dosing formula. A clinician may want to review CGM trends, meal patterns, insulin delivery, and hypoglycemia risk around the transition. Patients should not independently copy the trial's insulin changes or abruptly alter therapy based on this report.

The next useful study would be a larger randomized withdrawal trial with continued CGM follow-up. It should track insulin dose, weight, appetite, HbA1c, ketones, severe hypoglycemia, and diabetic ketoacidosis for longer than 12 weeks. Until then, this extension is a reason to monitor carefully, not a reason to overstate certainty.

Sources

1. Montaser E, et al. “Glycaemic Changes After Semaglutide Discontinuation in Adults With Type 1 Diabetes Using Automated Insulin Delivery: Analysis of the ADJUST-T1D Extension Study.” Diabetes, Obesity and Metabolism, 2026. https://pubmed.ncbi.nlm.nih.gov/42675550/

2. Shah VN, et al. “Semaglutide in Adults with Type 1 Diabetes and Obesity.” NEJM Evidence, 2025. https://pubmed.ncbi.nlm.nih.gov/40550013/

3. Karakus KE, et al. “Effect of Semaglutide on Insulin Dose Reduction in Adults With Type 1 Diabetes and Obesity Using Automated Insulin Delivery Systems: ADJUST-T1D Post Hoc Analysis.” Diabetes Care, 2026. https://pubmed.ncbi.nlm.nih.gov/41429002/

4. ADJUST-T1D trial registration, NCT05537233. https://clinicaltrials.gov/study/NCT05537233

Evidence grade: C+. The withdrawal signal comes from prospectively collected CGM data linked to a randomized parent trial, with multiplicity adjustment and a consistent sensitivity analysis. Confidence is limited by the small withdrawal group, nonrandomized extension comparison, and 12-week follow-up.

Medical disclaimer: This article is for educational purposes and is not medical advice. Semaglutide use in type 1 diabetes requires specialist oversight. Do not start, stop, or change semaglutide or insulin doses without guidance from a qualified clinician who can review your glucose data, medications, and individual risks. Oria exists to help readers understand peptide and metabolic research, not to replace diagnosis or treatment.