People with type 1 diabetes develop obesity too, and the drug class that treats obesity was developed, tested and approved in people whose pancreas still makes insulin. That mismatch is the whole subject. The approvals in this class cover type 2 diabetes, obesity, cardiovascular risk and liver disease — the first split is described in the two-approvals article — and none of them covers type 1.
What the labels say, and what they do not say
The labels are not uniformly silent, and the difference matters. The Wegovy label states that use of the drug in patients with type 1 diabetes or in combination with insulin has not been evaluated.[1] The Ozempic, Mounjaro and Zepbound labels do not mention type 1 diabetes at all.[2][3] One is a declared gap and the others are an absence. Neither is an authorization, and none of the four carries a type 1 indication.
A 2026 consensus report from a group of diabetes technology and endocrinology bodies states the position directly: regulatory approval of weekly GLP-1 and dual GLP-1/GIP receptor agonists in type 1 diabetes has not been achieved. It names hypoglycemia and hyperglycemia-related ketosis as the potential safety risks, and makes a point that belongs on a telehealth page: because the obesity indication already provides access, people with type 1 diabetes obtain these drugs without the type 1 specific safety education that would accompany a licensed use.[4] That is a description of the actual purchasing route, not a hypothetical one.
The adjunct trials found a real effect and a real signal
ADJUNCT ONE randomized 1,398 adults with type 1 diabetes three to one to liraglutide 1.8, 1.2 or 0.6 mg daily or placebo, added to treat-to-target insulin, for 52 weeks. From a mean baseline HbA1c of 8.2%, the estimated treatment difference against placebo was −0.20% at 1.8 mg (95% CI, −0.32 to −0.07) and −0.15% at 1.2 mg (95% CI, −0.27 to −0.03). Body weight fell by 4.9 kg more than placebo at 1.8 mg (95% CI, −5.7 to −4.2). Insulin dose fell too, at an estimated treatment ratio of 0.92 (95% CI, 0.88 to 0.96).[5]
Then the column that gets left out. Symptomatic hypoglycemia rose in every liraglutide group, at rate ratios of 1.31, 1.27 and 1.17 descending with dose. Hyperglycemia with ketosis rose significantly at 1.8 mg only, at an event rate ratio of 2.22 (95% CI, 1.13 to 4.34).[5] The dose with the largest weight and glycemic benefit is the dose carrying the ketosis signal, and the trial’s own conclusion says those safety findings limit clinical use in this group.
ADJUNCT TWO ran 835 patients for 26 weeks on capped insulin and split the two harms across different rungs. Symptomatic hypoglycemia was higher than placebo at 1.2 mg (21.3 against 16.6 events per patient-year, P = 0.03), while hyperglycemia with ketosis above 1.5 mmol/L was higher at 1.8 mg (0.5 against 0.1 events per patient-year, P = 0.01).[6]No rung was clean, and the two risks did not sit on the same rung.
Why reduced insulin is the sharp end
In type 2 diabetes an insulin reduction is a convenience. In type 1 it is the mechanism by which ketoacidosis becomes possible, because there is no endogenous reserve to cover the gap. A post hoc analysis of a modern semaglutide trial measured how large that reduction gets and where it comes from: over 26 weeks the total daily dose fell 22.6% (95% CI, −28.3 to −17.0), driven more by bolus insulin (−30.5%) than basal (−15.6%), with the basal share of the total rising from 0.56 to 0.62 (P < 0.001).[7]
The same analysis separates cause from consequence over time. At week 4, 83% of the reduction was a direct drug effect and 17% attributable to weight loss. By week 26 it was 52% against 48%.[7] So the common reassurance that the insulin drop is simply a consequence of losing weight is false early in treatment and only half true later, which is the period when a dose is being escalated and the titration schedule is still moving.
The DKA reputation belongs mostly to a different drug class
A 2026 systematic review of 90 studies of adjunct therapy in type 1 diabetes separated the two classes people conflate. For GLP-1 receptor agonists it found HbA1c −0.56%, weight −3.6 kg and BMI −1.05 kg/m², with severe hypoglycemia and DKA rare. For SGLT2 inhibitors it found a DKA relative risk of 2.19 (95% CI, 1.16 to 4.17), and no equivalent elevation for the GLP-1 class.[8]
A dedicated 2026 safety meta-analysis of 25 studies reached compatible numbers and graded them honestly. Overall hypoglycemia carried a relative risk of 1.01 and DKA a relative risk of 0.60 — but that DKA figure was rated very low certainty, the weakest grade in the paper.[9]What that review rated as moderate certainty was the tolerability penalty: nausea RR 2.89, vomiting RR 3.10, and early treatment withdrawal RR 2.02. Tolerability improved significantly after six months, so the same drug reads as intolerable at three months and manageable at twelve.[9]
A third pooled analysis, of 13 studies in adults with type 1 diabetes and a BMI of 25 or above, shows the two tails of the glucose distribution moving in opposite directions at once: hypoglycemic episodes rose (OR 1.34; 95% CI, 1.02 to 1.76) while hyperglycemic episodes fell (OR 0.69; 95% CI, 0.56 to 0.87), alongside weight −4.31 kg, HbA1c −0.25% and insulin −9.24 units per day.[10] Whether this class looks safe in type 1 depends on which tail is counted.
A ketosis endpoint is not a DKA endpoint
The distinction has been observed directly. A double-blind crossover trial added weekly semaglutide to automated insulin delivery in 28 adults with type 1 diabetes. Time in range improved by 4.8 percentage points (P = 0.006) with no increase in time below range, and there were no cases of diabetic ketoacidosis and no severe hypoglycemia. There were two episodes of recurrent euglycemic ketosis without acidosis during semaglutide.[11]
Recorded against a DKA endpoint, that trial reports zero events. Recorded against a ketone measurement, it reports two. Both statements are true, and only the second tells a reader with type 1 diabetes why a ketone meter is part of the conversation — as is what happens during an intercurrent illness, covered for the general case in the acute illness article.
What the newest randomized evidence shows
ADJUST-T1D randomized 72 adults with type 1 diabetes, a BMI of 30 or above and an automated insulin delivery system to semaglutide up to 1 mg weekly or placebo for 26 weeks. A composite of time in range above 70%, time below range under 4% and at least 5% weight loss was met by 36% against 0%, a difference of 36 percentage points (95% CI, 20.6 to 52.2). Weight fell 8.8 kg (95% CI, −10.6 to −7.0), time in range rose 8.8 points, and severe hypoglycemia occurred twice in each group with no DKA reported.[12]
That is 72 people for six months. Against it sits the largest observational dataset in this literature: 174,678 patients with type 1 diabetes in a national record, analyzed as a sequence of emulated trials. Five-year risk of major adverse cardiovascular events was 4.3% against 5.0% (hazard ratio 0.85; 95% CI, 0.77 to 0.95) and end-stage kidney disease 1.6% against 1.9% (hazard ratio 0.81; 95% CI, 0.69 to 0.95), with no increase in DKA hospitalization or severe hypoglycemia.[13] The kidney result in type 2 diabetes is in the FLOW article, and this is an observational echo of it in a population no trial has enrolled.
One more reversal worth keeping. A survey of 230 adults with type 1 diabetes using these drugs off-label found symptomatic hypoglycemia reported by 29% on tirzepatide against 13.4% on semaglutide (P < 0.001), with no difference in gastrointestinal events.[14] The agent that produces more weight loss produced roughly double the hypoglycemia reporting, which is the opposite of how the two are usually ranked on a tirzepatide price board.
What this does not settle
Nothing above makes a prescription appropriate or inappropriate for a particular person with type 1 diabetes. It establishes that the effect on weight and insulin requirement is real and reproducible, that the ketosis risk is dose-dependent and concentrated at the top of the ladder, and that the DKA alarm attached to adjunct therapy in general is mostly someone else’s finding.
The purchasing reading is narrower still. Compounded semaglutide and tirzepatide are not FDA-approved and are not reviewed by the FDA for safety, efficacy or quality before they are dispensed, so a compounded vial used in type 1 diabetes is an off-label use of an unapproved preparation — two separate distances from the evidence above, set out in the compounding article. An intake that never asks which type of diabetes an applicant has cannot route the question to anyone, which is one of the patterns in the telehealth article.