GLP-1 for Type 1 Diabetes: Off-Label Use With Insulin Pumps

About 1.6 million US adults live with type 1 diabetes, and a rising share also meet criteria for obesity — a combination clinicians sometimes call “double diabetes.” Insulin is non-negotiable, but the published evidence on adding a GLP-1 receptor agonist on top of insulin is now sizable: three liraglutide RCTs (ADJUNCT ONE^[1], ADJUNCT TWO^[2], Lira-1^[3]) and an early semaglutide case series from Buffalo (Dandona 2023 NEJM^[4]) describe a consistent pattern: modest HbA1c improvement, meaningful weight loss, a 10–20% drop in total daily insulin, and a manageable but real safety signal around ketones. The use is off-label in T1D, the protocol depends on a CGM and a responsive prescriber, and the modern hybrid closed-loop pumps (Tandem Control-IQ^[6], Medtronic 780G, Omnipod 5) make the workflow safer than it was a decade ago. This article walks through the trials, the dose-reduction rules, the DKA discussion, and the practical CGM-guided protocol most endocrinologists are now using.

The honest summary

• GLP-1 use in T1D is off-label. No GLP-1 receptor agonist carries an FDA indication for type 1 diabetes. The ADA 2025 Standards of Care^[9] note adjunctive use only in selected adults with T1D and obesity, under endocrinology supervision and with appropriate insulin-dose reductions.
• HbA1c reduction is modest. ADJUNCT ONE^[1] reported placebo-corrected HbA1c reductions of roughly −0.20% (liraglutide 1.2 mg) and −0.24% (1.8 mg) at week 52; Lira-1^[3] and Dandona^[4] describe similar magnitude. T1D is not where you go for big A1c wins from a GLP-1.
• Weight loss is meaningful. Across ADJUNCT ONE^[1], ADJUNCT TWO^[2], and Lira-1^[3], mean weight loss with liraglutide 1.8 mg in T1D was roughly 4–6 kg over 26–52 weeks — smaller in absolute terms than in T2D but proportional once BMI is matched.
• Insulin drops 10–20%. Total daily insulin fell ~10% on liraglutide 1.2 mg and ~13–18% on 1.8 mg in ADJUNCT ONE^[1]; basal and bolus fall together. That is why pre-emptive dose reduction at initiation is the safest practice.
• Ketone signal is real but manageable. ADJUNCT ONE^[1] reported a higher rate of symptomatic hyperglycemia with ketosis on liraglutide vs placebo, and ADJUNCT TWO^[2] confirmed the signal. This is not the euglycemic DKA epidemiology of SGLT2 inhibitors in T1D (Peters 2015^[8]), but it warrants beta-hydroxybutyrate checks during illness and a clear sick-day plan.

The patient population: T1D plus obesity is increasingly common

Modern T1D is no longer the lean, ketosis-prone phenotype of textbooks. The same obesogenic environment that doubled the prevalence of T2D has reached adults with T1D, and intensive insulin therapy itself promotes some weight gain. Roughly a third of US adults with T1D now meet criteria for obesity, which is why the off-label question keeps surfacing: insulin is mandatory, the lifestyle prescription is the same as for any obese adult, but the obesity-medicine toolkit was built for T2D. The ADA 2025 Standards^[9] are explicit that obesity in T1D should be treated, and that GLP-1 RAs may be considered as adjunctive therapy in selected patients under endocrinology supervision — a clear shift from the “not recommended” stance of a decade ago.

ADJUNCT ONE: the largest liraglutide-in-T1D RCT

ADJUNCT ONE (Mathieu 2016, Diabetes Care^[1]) randomized 1,398 adults with T1D and HbA1c 7.0–10.0% to liraglutide 1.8 mg, 1.2 mg, 0.6 mg, or placebo on top of treat-to-target insulin over 52 weeks. Placebo-corrected HbA1c reductions were −0.20% (1.8 mg), −0.15% (1.2 mg), and −0.09% (0.6 mg). Mean weight change was approximately −5.1 kg (1.8 mg), −4.0 kg (1.2 mg), and −2.5 kg (0.6 mg), vs +1.0 kg on placebo. Total daily insulin dose fell about 13% on 1.8 mg and 10% on 1.2 mg. The cost: liraglutide raised the rate of symptomatic hyperglycemia with ketosis and modestly increased confirmed-hypoglycemia events. The takeaway has held up — liraglutide works in T1D, but the metabolic gain comes with a ketone signal that must be monitored.

ADJUNCT TWO and Lira-1: replication with capped insulin

ADJUNCT TWO (Ahren 2016^[2]) ran the same dose comparison against placebo in 835 patients but kept the insulin dose capped at baseline, isolating the GLP-1 effect. At 26 weeks, HbA1c fell about 0.35% more on liraglutide 1.8 mg than on placebo, weight fell about 5 kg, and total daily insulin dropped roughly 8–10%. Lira-1 (Dejgaard 2016, Lancet Diabetes Endocrinol^[3]) randomized 100 adults with T1D and BMI ≥ 25 to liraglutide 1.2 mg or placebo on top of insulin for 24 weeks; HbA1c fell ~0.2% and weight fell ~6.8 kg net of placebo, with a basal insulin reduction of about 8 units/day. Across the three RCTs the picture is consistent: modest A1c, meaningful weight loss, clear insulin reduction, manageable ketone signal.

Semaglutide: the Buffalo case series and what came after

The most-cited semaglutide-in-T1D data is the Dandona NEJM 2023 letter^[4] reporting 10 adults with newly-diagnosed type 1 diabetes started on semaglutide alongside basal-bolus insulin. Within months, prandial insulin was eliminated in all 10, basal insulin was eliminated in 7, and HbA1c improved markedly — raising the question of whether GLP-1 RAs can preserve residual beta-cell function in the honeymoon window. The reply letter^[5] addressed methodological critiques and confirmed durability at one year in most patients. The Buffalo data is hypothesis- generating, not pivotal-trial-quality — the population is highly selected (new-onset, high residual C-peptide) and randomized trials are still in progress — but it is the reason semaglutide-in-T1D is being prescribed off-label in academic endocrinology clinics today.

Mechanism: glucagon suppression and postprandial control

GLP-1 receptor agonists work in T1D through the same three mechanisms as in T2D: glucagon suppression (most adults with T1D retain alpha-cell function and inappropriately secrete glucagon), delayed gastric emptying (which flattens postprandial glucose excursions), and central appetite suppression (which drives the weight and insulin-dose reduction). What is missing in T1D is the insulinotropic arm — there are essentially no beta cells to stimulate — which is exactly why the HbA1c effect is smaller than in T2D but the postprandial and weight effects survive. Khan 2024^[10] reviews the broader hypothesis that GLP-1 RAs may also delay T1D onset in high-risk pre-diabetic autoimmunity, which is the question the new randomized trials are designed to answer.

The DKA discussion: GLP-1 vs SGLT2 in T1D

Any conversation about off-label adjuncts in T1D has to address ketoacidosis. SGLT2 inhibitors in T1D carry a well-documented euglycemic DKA risk (Peters 2015^[8]) that is the reason the FDA has never approved them for T1D despite multiple sponsor submissions. The GLP-1 picture is meaningfully different. ADJUNCT ONE^[1] and ADJUNCT TWO^[2] reported a small but real excess of symptomatic hyperglycemia with ketosis on liraglutide vs placebo, but the absolute rate of clinically-adjudicated DKA was low and not statistically elevated. The mechanism is plausible — GLP-1-driven appetite suppression plus pre-emptive insulin reduction can leave a patient underinsulinized during illness — but the magnitude is an order of magnitude smaller than the SGLT2 signal. The practical implication is straightforward: GLP-1 in T1D requires a sick-day plan and a home beta-hydroxybutyrate meter, not the same level of fear that surrounds SGLT2 use in this population.

Magnitude: HbA1c change in published T1D trials

Hybrid closed-loop pumps make GLP-1 adjunct safer

The Brown 2019 NEJM RCT^[6] randomized 168 adults with T1D to closed-loop control (Tandem Control-IQ) or sensor-augmented pump for six months and reported a roughly 11-percentage-point improvement in time-in-range (70–180 mg/dL), driven mostly by less nocturnal hypoglycemia. The Breton 2021 real-world dataset^[7]showed the same effect held up at one year in routine use. For a T1D patient adding a GLP-1, that translates directly into safety margin: when meal-time gastric emptying is delayed by liraglutide or semaglutide, the closed-loop algorithm sees the resulting postprandial glucose curve and adjusts basal delivery automatically, blunting the hypoglycemia risk that comes with a fixed bolus calculator. Medtronic 780G and Omnipod 5 have not been compared head-to- head with GLP-1 add-on, but the directional benefit should be similar.

Practical initiation protocol

1. Endocrinology supervision is non-negotiable. Off-label use in T1D belongs with a prescriber who can read a CGM trace and respond to a sick-day call. Primary care or obesity-medicine-only management is not appropriate.
2. Reduce basal insulin 10–20% at initiation. The ADJUNCT-trial^[1][2] insulin adjustment ranges support a 10–15% basal reduction on day one of the 0.6 mg starting dose and another 5–10% when the dose escalates. Hybrid closed-loop users let the algorithm down-titrate.
3. Reduce meal-time bolus 10–20% for the first 4–6 weeks. Gastric emptying is delayed and carbohydrate absorption is slower than the patient's insulin-to-carb ratio expects; over-bolusing the first bite is the most common cause of postprandial hypoglycemia in this protocol.
4. Keep a CGM running. Dexcom G7, Libre 3, or the pump-integrated sensor — alerts at 70 and 180 mg/dL, urgent-low at 55. The CGM does most of the safety work.
5. Get a home beta-hydroxybutyrate meter. Precision Xtra or Keto-Mojo. Check BHB any time glucose is > 250 mg/dL for 2 hours, any illness with nausea, or any unexpected weight loss acceleration. Hold the next dose and contact the prescriber if BHB > 1.5 mmol/L.
6. Sick-day plan. During infection, fever, or inability to eat, do not stop insulin; do not skip basal; check BHB; hydrate; call the diabetes team early. The GLP-1 dose can be skipped that week.
7. Stop in pregnancy. GLP-1 RAs are not recommended in pregnancy; T1D management with intensive insulin and CGM takes priority. Discontinue at least 2 months before planned conception for semaglutide-based agents.

Insurance and access

Most US payers do not cover GLP-1 RAs for off-label use in T1D without a separate obesity diagnosis. Wegovy and Zepbound carry obesity indications and are the most likely to be approved when BMI ≥ 30 (or ≥ 27 with comorbidity); the comorbidity field can list T1D, but the indication on the prescription is obesity. Ozempic and Mounjaro carry T2D indications only and are typically not covered for T1D patients even with obesity; off-label prior-authorization attempts almost always fail. Compounded semaglutide via a 503A pharmacy is occasionally used but is now restricted following the FDA shortage-resolution declaration. Insulin pump and CGM coverage is separate and generally well established for T1D under Medicare and most commercial plans.

Related research and tools

• GLP-1 plus SGLT2 stacking — why SGLT2 in T1D carries a documented DKA risk (Peters 2015) and why the GLP-1 signal is smaller
• Trulicity vs Ozempic vs Mounjaro in T2D — the T2D head-to-head data that informs the off-label T1D dose-response projection
• GLP-1 and history of pancreatitis — another off-label-adjacent decision that turns on balancing trial data against a known risk signal
• GLP-1 side-effect timeline — week-by-week expectations for liraglutide and semaglutide initiation

Important disclaimer. This article is educational and does not constitute medical advice. GLP-1 receptor agonists are not FDA-approved for type 1 diabetes; any off-label use should be supervised by an endocrinologist familiar with CGM-guided insulin adjustment. Pre-emptive insulin reduction, home beta-hydroxybutyrate monitoring, and a written sick-day plan are non-negotiable elements of safe use. Pregnancy is a contraindication; discontinue GLP-1 therapy at least two months before planned conception. Insurance coverage for off-label use is rare; an obesity indication is usually required. PMIDs were verified live against the PubMed E-utilities API on 2026-05-29.

Last verified: 2026-05-29. Next review: every 12 months, or sooner if new randomized data on semaglutide or tirzepatide adjunct in type 1 diabetes is published.