Liraglutide

What it is

Liraglutide is a man-made analogue of glucagon-like peptide-1, a hormone the gut releases after a meal. It is 97% identical to human GLP-1(7-37), with an added fatty-acid side chain that lets it bind to albumin in the blood so one injection lasts a day. It is given under the skin once daily and is sold under two brand names at two different dose ranges: Victoza, up to 1.8 mg a day for type 2 diabetes, and Saxenda, up to 3 mg a day for weight management.

What the research says

For type 2 diabetes the evidence is strong and includes hard outcomes. In LEADER, 9,340 people at high cardiovascular risk followed for a median 3.8 years, the composite of cardiovascular death, non-fatal heart attack and non-fatal stroke occurred in 13.0% on liraglutide versus 14.9% on placebo, and cardiovascular death in 4.7% versus 6.0%. For weight the effect is real and moderate: 8.4 kg versus 2.8 kg at 56 weeks in people without diabetes, 6.0% versus 2.0% of body weight in people with diabetes. Off-label it resolved steatohepatitis in a 52-patient phase 2 trial, reduces alcohol consumption in pooled studies of its class, and flatly failed in advanced heart failure. The harms are mostly gastrointestinal and common: nausea in 39.3% versus 13.8% on placebo in the weight trials. Rarer signals are gallbladder disease (pooled relative risk 1.37 across 76 trials, 2.29 in weight-loss trials), pancreatitis (pooled relative risk 1.44), and a rodent thyroid C-cell tumour finding that drives the boxed warning and remains unresolved in humans. Stopping is followed by weight regain of about 0.4 kg a month.

Evidence grade: Well established.

How it works

Drug class: Glucagon-like peptide-1 (GLP-1) receptor agonist (incretin mimetic), given by daily subcutaneous injection

Liraglutide is a modified copy of the human gut hormone GLP-1, sharing 97% of its amino acid sequence, with a fatty-acid tail that makes it last about a day instead of minutes. It switches on GLP-1 receptors on the insulin-producing beta cells, so they release more insulin but only when blood glucose is already high, which is why it does not cause hypoglycaemia on its own. It also reduces the glucagon that raises blood glucose, and slows the stomach from emptying. The slowed stomach and effects on appetite signalling in the brain are what produce the weight loss and also the nausea. (Source 1)

Boxed warning

WARNING: RISK OF THYROID C-CELL TUMORS • Liraglutide causes dose-dependent and treatment-duration-dependent thyroid C-cell tumors at clinically relevant exposures in both genders of rats and mice. It is unknown whether VICTOZA causes thyroid C-cell tumors, including medullary thyroid carcinoma (MTC), in humans, as the human relevance of liraglutide-induced rodent thyroid C-cell tumors has not been determined [see Warnings and Precautions ( 5.1 ), Nonclinical Toxicology ( 13.1 )]. • VICTOZA is contraindicated in patients with a personal or family history of MTC and in patients with Multiple Endocrine Neoplasia syndrome type 2 (MEN 2). Counsel patients regarding the potential risk for MTC with the use of VICTOZA and inform them of symptoms of thyroid tumors (e.g., a mass in the neck, dysphagia, dyspnea, persistent hoarseness). Routine monitoring of serum calcitonin or using thyroid ultrasound is of uncertain value for early detection of MTC in patients treated with VICTOZA [see Contraindications ( 4 ), Warnings and Precautions ( 5.1 )].

(Source 2)

What it is used for

• Licensed for adults and for children aged 10 and over, at up to 1.8 mg daily. The effect on HbA1c compared with other glucose-lowering drugs is modest - about 0.26 of a percentage point in a 25-trial meta-analysis - with 2 to 3 kg of weight loss alongside it. Evidence: established. (Source 3)
• This is the strongest evidence liraglutide has. In LEADER, 9,340 people followed a median 3.8 years, the composite of cardiovascular death, non-fatal heart attack and non-fatal stroke fell from 14.9% to 13.0%, an absolute reduction of 1.9 percentage points, and cardiovascular death from 6.0% to 4.7%. Note that the individual components other than death were not significant on their own. Evidence: established. (Source 4)
• At 3 mg daily, 5.6 kg more weight lost than placebo at 56 weeks in people without diabetes, and 4 percentage points of body weight more in people with type 2 diabetes. Both are weight outcomes over about a year, not outcomes like heart attacks or death, and the weight returns at about 0.4 kg a month after stopping. Evidence: established. (Source 5)
• One 52-patient phase 2 trial found steatohepatitis resolved on biopsy in 39% on liraglutide versus 9% on placebo, with less fibrosis progression. That is a single small trial with a confidence interval whose lower bound touches 1.0, and its own authors called for extensive longer-term studies. Evidence: limited. (Source 6)
• The FIGHT trial in 300 patients recently hospitalised with heart failure found no improvement in clinical stability, and deaths and readmissions were numerically higher on liraglutide. The investigators concluded the findings do not support this use. Evidence: not-supported. (Source 7)
• A 2025 systematic review pooling 14 studies, only four of them randomised, found a 7.81-point drop in AUDIT scores with GLP-1 receptor agonists as a class. Heterogeneity was very high at 87.5% and most of the signal comes from observational data and from semaglutide rather than liraglutide, so this is promising but not established. Evidence: limited. (Source 8)

Interactions

• Insulin and sulfonylureas (insulin secretagogues) (label): Liraglutide on its own rarely causes low blood sugar, but added to insulin or a sulfonylurea it can, including severely; the label advises considering a dose reduction of those drugs. (Source 9)
• Any medicine or supplement swallowed by mouth (label): Liraglutide slows the stomach emptying, so anything taken orally can be absorbed later or less completely. In the formal studies this did not matter clinically for the drugs tested, but the label still asks for caution, and the same mechanism applies to oral supplements even though none were tested. (Source 9)
• Digoxin (pharmacokinetic study): A formal interaction study found liraglutide reduced digoxin exposure by about a sixth and its peak level by nearly a third, and delayed the peak. (Source 10)
• Atorvastatin (pharmacokinetic study): Total exposure to atorvastatin was unchanged but its peak concentration dropped by 38% and arrived two hours later - a timing effect rather than a loss of drug. (Source 10)
• Paracetamol (acetaminophen) (pharmacokinetic study): Overall exposure unchanged, peak concentration 31% lower - worth knowing for anyone relying on a single dose for fast pain relief. (Source 10)
• Combined oral contraceptives (pharmacokinetic study): Peak levels of both hormones fell by about an eighth and levonorgestrel total exposure rose 18%; overall oestrogen exposure was unchanged. (Source 11)
• Alcohol (clinical trial): There is no pharmacokinetic interaction study and no warning about alcohol in the labels we read. What the literature documents is the reverse direction: GLP-1 receptor agonists, liraglutide among them, are associated with drinking less. A 2025 systematic review pooled a 7.81-point fall in AUDIT scores, mostly from observational studies (10 of 14) with very high heterogeneity. The authors say large, dedicated trials are still needed to confirm this. (Source 8)
• Food and meal timing (label): No food restriction applies to the injection itself; the label says it can be given at any time of day, with or without meals. The food-related effect runs the other way, through delayed stomach emptying and reduced appetite. (Source 12)
• Oral supplements taken for absorption-sensitive reasons (for example levothyroxine-style narrow-window products, iron, calcium) (theoretical): No supplement was tested in the interaction programme. The documented mechanism - slowed stomach emptying - applies to anything swallowed, so the expectation is delayed rather than reduced absorption; this is extrapolation from the label's own statement, not measured data. (Source 9)

Stopping it

• Liraglutide does not produce a withdrawal syndrome, and nothing in the sources we read describes dependence or a taper. What stopping does produce is weight regain. A 2026 BMJ systematic review and meta-analysis of 37 studies found weight came back at about 0.4 kg a month after weight-management drugs were stopped, faster than after behavioural programmes, and independent of how much had been lost. (Source 13)
• The metabolic improvements go too. The same review projected that every cardiometabolic marker it tracked would return to its starting value within about eighteen months of stopping. (Source 13)
• The label builds planned stopping into the weight indication rather than treating it as failure: it tells prescribers to check weight at 16 weeks and to stop if the person has not lost at least 4% of their starting weight. (Source 14)
• Dose escalation is itself a form of managed starting and stopping, driven by nausea: the label's stepped schedule exists to limit gastrointestinal reactions, and it tells prescribers to stop the drug if the target dose cannot be tolerated. (Source 14)

What goes wrong

In the same heart failure trial deaths and heart failure readmissions were numerically more frequent on liraglutide, though the differences were not statistically significant. (Source 7)

• Randomized trial, Low certainty.
• Size: 300 patients.
• Who: Adults recently hospitalised with heart failure and reduced ejection fraction.
• How long: 180 days.
• Result: Deaths 19 (12%) vs 16 (11%), hazard ratio 1.10 (95% CI 0.57 to 2.14), P = .78; heart failure rehospitalisation 63 (41%) vs 50 (34%), hazard ratio 1.30 (95% CI 0.89 to 1.88), P = .17.
• Funding: not stated in the record we read; the trial was not powered for these outcomes, so this is a signal and not a demonstrated harm.

There were no significant between-group differences in the number of deaths (19 [12%] in the liraglutide group vs 16 [11%] in the placebo group; hazard ratio, 1.10 [95% CI, 0.57-2.14]; P = .78) or rehospitalizations for heart failure (63 [41%] vs 50 [34%], respectively; hazard ratio, 1.30 [95% CI, 0.89-1.88]; P = .17) or for the exploratory secondary end points.

In that same steatohepatitis trial, gastrointestinal side effects were the one adverse-event class clearly more common on liraglutide. (Source 6)

• Randomized trial, Low certainty.
• Size: 45 patients with adverse event data.
• Who: Overweight patients with non-alcoholic steatohepatitis.
• How long: 48 weeks.
• Result: Gastrointestinal disorders in 21/23 (81%) vs 17/22 (65%): diarrhoea 10 (38%) vs 5 (19%), constipation 7 (27%) vs none, loss of appetite 8 (31%) vs 2 (8%)
• Funding: Wellcome Trust, National Institute of Health Research, and Novo Nordisk.

with the exception of gastrointestinal disorders in 21 (81%) of 23 patients in the liraglutide group and 17 (65%) of 22 patients in the placebo group, which included diarrhoea (ten [38%] patients in the liraglutide group vs five [19%] in the placebo group), constipation (seven [27%] vs none), and loss of appetite (eight [31%] vs two [8%]).

Across 76 randomised trials of GLP-1 receptor agonists, the class raised the risk of gallbladder and biliary disease, and the risk more than doubled in trials of weight loss. (Source 15)

• Meta-analysis, Moderate certainty.
• Size: 76 randomised trials, 103,371 patients; mean age 57.8 years, 40.5% women.
• Who: Adults in GLP-1 receptor agonist trials against placebo or non-GLP-1 drugs, for diabetes and for weight loss.
• How long: Varied by trial.
• Result: Gallbladder or biliary disease relative risk 1.37 (95% CI 1.23-1.52); cholelithiasis 1.27 (1.10-1.47); cholecystitis 1.36 (1.14-1.62); biliary disease 1.55 (1.08-2.22). In weight-loss trials (n=13) relative risk 2.29 (1.64-3.18) vs 1.27 (1.14-1.43) in diabetes and other trials, P<.001 for interaction; higher doses relative risk 1.56 (1.36-1.78)
• Funding: not stated in the record we read; GRADE used to rate each outcome.

Among all included trials, randomization to GLP-1 RA treatment was associated with increased risks of gallbladder or biliary diseases (RR, 1.37; 95% CI, 1.23-1.52); specifically, cholelithiasis (RR, 1.27; 95% CI, 1.10-1.47), cholecystitis (RR, 1.36; 95% CI, 1.14-1.62), and biliary disease (RR, 1.55; 95% CI, 1.08-2.22).

Pooled randomised trial data show a modestly increased risk of pancreatitis with GLP-1 receptor agonists, although the signal does not hold up in the subgroup analyses. (Source 16)

• Meta-analysis, Low certainty.
• Size: 62 studies, 66,232 patients; mean age 58.3 years.
• Who: Adults in randomised trials of dulaglutide, exenatide, liraglutide, semaglutide, beinaglutide, retatrutide or tirzepatide.
• How long: Mean follow-up 43.5 weeks (range 1 to 198)
• Result: Pancreatitis relative risk 1.44 (95% CI 1.09-1.89), p = 0.009; with background medications 1.28 (0.87-1.87) and without 1.37 (0.91-2.05), neither significant. Pancreatic cancer 1.30 (0.86-1.97) overall.
• Funding: not stated in the record we read; the authors note many studies were excluded because both arms had zero events.

Limit of this finding: This review's numbers do not sit comfortably together. The overall pancreatitis estimate (relative risk 1.44) is higher than both of its subgroup estimates (1.28 and 1.37), which is unusual for a pooled result and is not explained in the abstract, and the abstract reports a significant pancreatic-cancer association in one subgroup (relative risk 1.85) while its conclusion calls the cancer finding slight. The authors also left out every trial in which neither group had a case, and say the difference they saw is likely minimal. Read this as a possible small signal that needs confirming, not as a measured size of risk.

Meta-analysis showed a significantly increased risk of pancreatitis (RR: 1.44, 95% CI 1.09-1.89, p = 0.009), but not when stratified by background medications (RR: 1.28, 95% CI 0.87-1.87) and without background medications (RR: 1.37, 95% CI 0.91-2.05).

In a French national nested case-control study, one to three years of GLP-1 receptor agonist use was associated with a higher rate of thyroid cancer, including medullary thyroid cancer - the tumour type the boxed warning concerns. (Source 17)

• Case-control study, Low certainty.
• Size: 2,562 thyroid cancer cases matched to 45,184 controls.
• Who: People with type 2 diabetes on second-line diabetes drugs in the French national health insurance database, 2006-2018, cancers identified 2014-2018.
• How long: Exposure measured over the 6 years before a 6-month lag period.
• Result: All thyroid cancer adjusted hazard ratio 1.58 (95% CI 1.27-1.95) for 1-3 years of use; medullary thyroid cancer adjusted hazard ratio 1.78 (95% CI 1.04-3.05)
• Funding: not stated in the record we read; observational design, so detection bias and residual confounding cannot be excluded and this is an association rather than a demonstrated cause.

Use of GLP-1 RA for 1-3 years was associated with increased risk of all thyroid cancer (adjusted hazard ratio [HR] 1.58, 95% CI 1.27-1.95) and medullary thyroid cancer (adjusted HR 1.78, 95% CI 1.04-3.05).

At the 3 mg weight-management dose, nausea affected 39.3% of patients versus 13.8% on placebo, and vomiting 15.7% versus 3.9%. (Source 18)

• Official position, Certainty not rated.
• Size: 3,384 patients on liraglutide and 1,941 on placebo across five double-blind placebo-controlled trials.
• Who: Adults with overweight or obesity, mean treatment duration 46 weeks.
• How long: Up to 56 weeks (one cohort to 160 weeks)
• Result: Placebo vs liraglutide 3 mg: nausea 13.8% vs 39.3%, diarrhoea 9.9% vs 20.9%, constipation 8.5% vs 19.4%, vomiting 3.9% vs 15.7%, dyspepsia 2.7% vs 9.6%, increased lipase 2.2% vs 5.3%.
• Funding: Manufacturer-sponsored registration trials reported in the label held by Novo Nordisk; SPL version 22, published 15 June 2026.

Nausea 13.8 39.3 Diarrhea 9.9 20.9 Constipation 8.5 19.4 Vomiting 3.9 15.7

Gastrointestinal side effects made more than twice as many people stop liraglutide as placebo in the weight-management trials. (Source 18)

• Official position, Certainty not rated.
• Size: 3,384 patients on liraglutide and 1,941 on placebo.
• Who: Adults with overweight or obesity in five placebo-controlled trials.
• How long: Up to 56 weeks.
• Result: Premature discontinuation for adverse reactions 9.8% vs 4.3%; nausea 2.9% vs 0.2%, vomiting 1.7% vs less than 0.1%, diarrhoea 1.4% vs 0%.
• Funding: Manufacturer-sponsored trials reported in the label held by Novo Nordisk.

In adult clinical trials, 9.8% of patients treated with SAXENDA and 4.3% of patients treated with placebo prematurely discontinued treatment as a result of adverse reactions.

At the lower diabetes doses the gastrointestinal effects are less frequent but still several times placebo. (Source 19)

• Official position, Certainty not rated.
• Size: 661 placebo, 645 on 1.2 mg, 1,024 on 1.8 mg, pooled from glycaemic-control trials.
• Who: Adults with type 2 diabetes, mean HbA1c 8.4%, average 9 years of diabetes.
• How long: Varied by trial.
• Result: Placebo vs 1.2 mg vs 1.8 mg: nausea 5% / 18% / 20%, diarrhoea 4% / 10% / 12%, vomiting 2% / 6% / 9%, decreased appetite 1% / 10% / 9%, dyspepsia 1% / 4% / 7%, constipation 1% / 5% / 5%.
• Funding: Manufacturer-sponsored trials reported in the label held by Novo Nordisk; SPL version 31, published 17 November 2025.

Nausea 5 18 20 Diarrhea 4 10 12 Headache 7 11 10 Nasopharyngitis 8 9 10 Vomiting 2 6 9 Decreased appetite 1 10 9

Liraglutide causes thyroid C-cell tumours in rats and mice in a dose- and duration-dependent way, and whether this happens in humans is still unknown; this is the basis of the boxed warning. (Source 2)

• Official position, Certainty not rated.
• Size: Not applicable - regulatory position based on rodent carcinogenicity studies.
• Who: Rats and mice; human relevance undetermined.
• How long: Position current as of SPL version 31, published 17 November 2025.
• Result: No human rate given; the label records that the drug is contraindicated in anyone with a personal or family history of medullary thyroid carcinoma or MEN 2, and that calcitonin monitoring and thyroid ultrasound are of uncertain value.
• Funding: Label held by Novo Nordisk.

Liraglutide causes dose-dependent and treatment-duration-dependent thyroid C-cell tumors at clinically relevant exposures in both genders of rats and mice.

Weight comes back after the drug stops, at roughly 0.4 kg a month, and the metabolic gains are projected to disappear within about eighteen months. (Source 13)

• Meta-analysis, Moderate certainty.
• Size: 37 studies, 63 intervention arms, 9,341 participants.
• Who: Adults with overweight or obesity who had taken a weight-management medication for at least 8 weeks and were followed at least 4 weeks after stopping.
• How long: Average treatment 39 weeks (range 11-176); average follow-up 32 weeks (4-104)
• Result: Average monthly weight regain 0.4 kg (95% CI 0.3 to 0.5), and 0.3 kg (0.2 to 0.4) per month versus control in randomised trials; regain faster after medication than after behavioural programmes by 0.3 kg (0.22 to 0.34) per month; all cardiometabolic markers projected to return to baseline within 1.4 years.
• Funding: not stated in the record we read; includes non-randomised and observational studies alongside randomised trials.

The average monthly rate of weight regain was 0.4 kg (95% confidence interval (CI) 0.3 to 0.5) (mixed model 0.3 kg (0.2 to 0.4) monthly v control in RCTs).

What the evidence supports

In people with type 2 diabetes and high cardiovascular risk, liraglutide reduced major adverse cardiovascular events by 1.9 percentage points in absolute terms over a median 3.8 years. (Source 4)

• Randomized trial, High certainty.
• Size: 9,340 patients (4,668 liraglutide, 4,672 placebo)
• Who: Adults with type 2 diabetes and high cardiovascular risk.
• How long: Median follow-up 3.8 years.
• Result: Primary composite in 608/4,668 (13.0%) vs 694/4,672 (14.9%), absolute difference 1.9 percentage points, about 53 people treated for 3.8 years per event avoided; hazard ratio 0.87 (95% CI 0.78 to 0.97), P<0.001 for non-inferiority, P=0.01 for superiority.
• Funding: Novo Nordisk and the National Institutes of Health.

The primary outcome occurred in significantly fewer patients in the liraglutide group (608 of 4668 patients [13.0%]) than in the placebo group (694 of 4672 [14.9%]) (hazard ratio, 0.87; 95% confidence interval [CI], 0.78 to 0.97; P<0.001 for noninferiority; P=0.01 for superiority).

In the same trial, deaths from cardiovascular causes and from any cause were both lower on liraglutide. (Source 4)

• Randomized trial, High certainty.
• Size: 9,340 patients.
• Who: Adults with type 2 diabetes and high cardiovascular risk.
• How long: Median 3.8 years.
• Result: Cardiovascular death 219 (4.7%) vs 278 (6.0%), hazard ratio 0.78 (95% CI 0.66 to 0.93), P=0.007, absolute difference 1.3 percentage points; all-cause death 381 (8.2%) vs 447 (9.6%), hazard ratio 0.85 (95% CI 0.74 to 0.97), P=0.02.
• Funding: Novo Nordisk and the National Institutes of Health.

Fewer patients died from cardiovascular causes in the liraglutide group (219 patients [4.7%]) than in the placebo group (278 [6.0%]) (hazard ratio, 0.78; 95% CI, 0.66 to 0.93; P=0.007).

At 3 mg daily in people with obesity but without diabetes, liraglutide produced 5.6 kg more weight loss than placebo at 56 weeks. (Source 5)

• Randomized trial, High certainty.
• Size: 3,731 patients (2,487 liraglutide, 1,244 placebo)
• Who: Adults without type 2 diabetes with BMI at least 30, or at least 27 with dyslipidaemia or hypertension; 78.5% women, 61.2% with prediabetes.
• How long: 56 weeks.
• Result: Weight change -8.4 +/- 7.3 kg vs -2.8 +/- 6.5 kg, difference -5.6 kg (95% CI -6.0 to -5.1), P<0.001; at least 5% lost by 63.2% vs 27.1%; more than 10% lost by 33.1% vs 10.6%.
• Funding: Novo Nordisk.

At week 56, patients in the liraglutide group had lost a mean of 8.4±7.3 kg of body weight, and those in the placebo group had lost a mean of 2.8±6.5 kg (a difference of -5.6 kg; 95% confidence interval, -6.0 to -5.1; P<0.001, with last-observation-carried-forward imputation).

In people who have type 2 diabetes as well as overweight or obesity, the weight effect of liraglutide 3.0 mg was smaller: 4 percentage points of body weight more than placebo. (Source 20)

• Randomized trial, High certainty.
• Size: 846 randomised (423 at 3.0 mg, 211 at 1.8 mg, 212 placebo)
• Who: Adults with BMI 27 or more, HbA1c 7.0% to 10.0%, on up to three oral diabetes drugs, at 126 sites in 9 countries.
• How long: 56 weeks plus 12-week off-drug follow-up.
• Result: Weight loss 6.0% (6.4 kg) at 3.0 mg, 4.7% (5.0 kg) at 1.8 mg, 2.0% (2.2 kg) on placebo; difference for 3.0 mg vs placebo -4.00% (95% CI -5.10 to -2.90), P<.001. At least 5% lost by 54.3% vs 21.4%, difference 32.9 percentage points (95% CI 24.6 to 41.2)
• Funding: not stated in the record we read; conducted by the SCALE programme sponsor.

Weight loss was 6.0% (6.4 kg) with liraglutide (3.0-mg dose), 4.7% (5.0 kg) with liraglutide (1.8-mg dose), and 2.0% (2.2 kg) with placebo (estimated difference for liraglutide [3.0 mg] vs placebo, -4.00% [95% CI, -5.10% to -2.90%]; liraglutide [1.8 mg] vs placebo, -2.71% [95% CI, -4.00% to -1.42%]; P < .001 for both).

Pooling 25 randomised trials, the glucose-lowering effect of liraglutide at diabetes doses relative to other glucose-lowering drugs is modest - about a quarter of an HbA1c percentage point. (Source 3)

• Meta-analysis, Moderate certainty.
• Size: 25 randomised trials, 10,593 participants.
• Who: Adults with type 2 diabetes and obesity.
• How long: 24 to 52 weeks.
• Result: Liraglutide 1.2 mg reduced HbA1c by -0.24% versus oral antidiabetic drugs; liraglutide 1.8 mg reduced HbA1c by -0.26% versus other glucose-lowering therapies; weight -2.30 kg (1.8 mg) and -1.24 kg (1.2 mg) versus other glucose-lowering therapies.
• Funding: not stated in the record we read; comparators were active drugs rather than placebo in most trials.

For glycemic control, liraglutide 1.2 mg reduced HbA1c by -0.24% versus OADs, while liraglutide 1.8 mg reduced HbA1c by -0.26% versus GLTs.

In a 52-patient phase 2 trial, liraglutide 1.8 mg resolved steatohepatitis on biopsy in 39% of patients versus 9% on placebo. (Source 6)

• Randomized trial, Low certainty.
• Size: 52 randomised (26 per arm); 45 had an end-of-treatment biopsy.
• Who: Overweight patients with clinical evidence of non-alcoholic steatohepatitis at four UK centres.
• How long: 48 weeks.
• Result: Resolution of definite NASH without worsening fibrosis in 9/23 (39%) vs 2/22 (9%), relative risk 4.3 (95% CI 1.0-17.7), p=0.019; fibrosis progression in 2/23 (9%) vs 8/22 (36%), 0.2 (0.1-1.0), p=0.04.
• Funding: Wellcome Trust, National Institute of Health Research, and Novo Nordisk; small phase 2 trial with a single-group design feature and a confidence interval whose lower bound touches 1.0.

Nine (39%) of 23 patients who received liraglutide and underwent end-of-treatment liver biopsy had resolution of definite non-alcoholic steatohepatitis compared with two (9%) of 22 such patients in the placebo group (relative risk 4·3 [95% CI 1·0-17·7]; p=0·019).

Pooled evidence suggests GLP-1 receptor agonists including liraglutide reduce alcohol consumption, but the pooling is dominated by observational data and is very heterogeneous. (Source 8)

• Meta-analysis, Very low certainty.
• Size: 14 studies (4 randomised trials, 10 observational), n = 5,262,268.
• Who: Adults with hazardous drinking or alcohol use disorder.
• How long: Varied.
• Result: AUDIT score mean difference -7.81 points (95% CI -9.02 to -6.60), I2 = 87.5%; reduced drinking days, units per drinking day and cravings reported particularly with semaglutide.
• Funding: No funding was received for this study, as stated by the authors; the very high heterogeneity and predominance of observational data make this an association rather than an established effect.

Limit of this finding: Only 4 of the 14 studies pooled were randomised trials; the other 10 were observational, so the drop in drinking scores cannot be put down to the drug alone. The review's own conclusion is that large, dedicated randomised trials are needed to confirm that these drugs help with alcohol use disorder and to define how they should be used, so this is a promising signal and not an established benefit.

Pooled analysis demonstrated a significant reduction in AUDIT scores (mean difference -7.81 points; 95% CI -9.02 to -6.60; I2 = 87.5%).

What the evidence does not support

The individual components of the LEADER composite - heart attack, stroke and heart failure admission - were not significantly reduced on their own. (Source 4)

• Randomized trial, High certainty.
• Size: 9,340 patients.
• Who: Adults with type 2 diabetes and high cardiovascular risk.
• How long: Median 3.8 years.
• Result: Rates numerically lower but not statistically significant; the trial states no adjustment for multiplicity was made for prespecified exploratory outcomes.
• Funding: Novo Nordisk and the National Institutes of Health.

The rates of nonfatal myocardial infarction, nonfatal stroke, and hospitalization for heart failure were nonsignificantly lower in the liraglutide group than in the placebo group.

In advanced heart failure with reduced ejection fraction, liraglutide did not improve clinical stability after a heart failure admission. (Source 7)

• Randomized trial, Moderate certainty.
• Size: 300 patients (154 liraglutide, 146 placebo)
• Who: Adults recently hospitalised with heart failure and reduced ejection fraction (median 25%) at 24 US sites; 59% had type 2 diabetes.
• How long: 180 days.
• Result: Mean global rank score 146 (liraglutide) vs 156 (placebo), P = .31; no significant difference in any exploratory secondary outcome.
• Funding: not stated in the record we read; phase 2 investigator-led trial.

Compared with placebo, liraglutide had no significant effect on the primary end point (mean rank of 146 for the liraglutide group vs 156 for the placebo group, P = .31).

Randomised trial evidence does not show a significant increase in thyroid cancer with this class of drug, but the events are too rare for the trials to rule one out. (Source 21)

• Meta-analysis, Very low certainty.
• Size: 15 randomised trials, 84,237 participants; 28 thyroid cancer events on drug vs 15 on control.
• Who: Adults in trials of approved incretin-based therapies with at least 26 weeks of follow-up.
• How long: At least 26 weeks per trial.
• Result: Odds ratio 1.52 (95% CI 0.86-2.68), I2 = 0.0%; certainty rated very low for serious imprecision.
• Funding: not stated in the record we read.

Meta-analysis showed no statistically significant association between incretin-based therapy use and thyroid cancer risk (odds ratio 1.52, 95% CI 0.86-2.68; I2 = 0.0%).

Where the research disagrees

Whether liraglutide raises the risk of thyroid cancer in humans, as it does in rodents

• The US prescribing information (boxed warning), Rodent carcinogenicity studies; a regulatory position, current as of 17 November 2025: Liraglutide causes dose-dependent and treatment-duration-dependent thyroid C-cell tumors at clinically relevant exposures in both genders of rats and mice. It is unknown whether VICTOZA causes thyroid C-cell tumors, including medullary thyroid carcinoma (MTC), in humans, as the human relevance of liraglutide-induced rodent thyroid C-cell tumors has not been determined (Source 2)
• Bezin and colleagues, French national health insurance data, Nested case-control study, 2,562 cases and 45,184 matched controls: In the current study we found increased risk of all thyroid cancer and medullary thyroid cancer with use of GLP-1 RA, in particular after 1-3 years of treatment. (Source 17)
• A 2026 meta-analysis restricted to randomised trials, Meta-analysis of 15 randomised trials, 84,237 participants, 43 events in total; certainty rated very low: RCT evidence does not demonstrate a statistically significant increase in thyroid cancer risk with incretin-based therapies. However, limited follow-up and imprecision prevent exclusion of a clinically relevant increase, supporting the need for long-term surveillance. (Source 21)

How much

• Reference intake: There is no reference intake for a prescription injectable; the dose is set by the prescriber. As a position, the US prescribing information for Victoza (SPL version 31, 17 November 2025) starts adults at 0.6 mg once daily for one week, which it states is not effective for glycaemic control and exists only to reduce gastrointestinal reactions, then 1.2 mg, then 1.8 mg if needed. Saxenda (SPL version 22, 15 June 2026) escalates weekly through 0.6, 1.2, 1.8 and 2.4 mg to 3 mg from week 5. (Source 22)
• Upper limit: No tolerable upper intake level exists for a prescription drug. As a position, the maximum recommended dosage of Victoza for type 2 diabetes is 1.8 mg once daily; for Saxenda in weight management the recommended adult dosage is 3 mg daily, with lower doses used only for titration. (Source 22)
• Studied: LEADER randomised 9,340 adults with type 2 diabetes and high cardiovascular risk to liraglutide or placebo added to standard care, with a median follow-up of 3.8 years. (Source 4)
• Studied: SCALE Obesity and Prediabetes gave once-daily subcutaneous liraglutide 3.0 mg or placebo to 3,731 adults for 56 weeks alongside lifestyle counselling. (Source 5)
• Studied: SCALE Diabetes compared once-daily subcutaneous liraglutide 3.0 mg (n = 423), 1.8 mg (n = 211) and placebo (n = 212) for 56 weeks with a 500 kcal/day dietary deficit and at least 150 minutes a week of activity. (Source 20)
• Studied: The LEAN steatohepatitis trial used subcutaneous liraglutide 1.8 mg daily against placebo for 48 weeks in 52 patients. (Source 6)
• Studied: FIGHT advanced liraglutide to 1.8 mg a day over the first 30 days as tolerated and continued it for 180 days in 300 patients with heart failure. (Source 7)

A common belief, and what the research shows

The belief: Once the weight is off with liraglutide, you can stop the injections and keep the result.

What the research shows: The pooled evidence says otherwise. A 2026 BMJ systematic review of 37 studies found that "The average monthly rate of weight regain was 0.4 kg (95% confidence interval (CI) 0.3 to 0.5) (mixed model 0.3 kg (0.2 to 0.4) monthly v control in RCTs)." The metabolic benefit goes with it: "All cardiometabolic markers were projected to return to baseline within 1.4 years after the cessation of WMM." Regain was also faster after stopping a drug than after stopping a behavioural programme, and did not depend on how much weight had been lost.

Questions and answers

What is it?

Liraglutide is a manufactured copy of GLP-1, a hormone your gut releases after eating. It matches human GLP-1 in 97% of its amino acid sequence and carries an added fatty-acid tail so that one injection under the skin lasts about a day instead of minutes. It is sold as Victoza for type 2 diabetes at up to 1.8 mg daily and as Saxenda for weight management at up to 3 mg daily. (Source 1)

What does it do in the body?

It switches on GLP-1 receptors on the pancreas's beta cells, raising cyclic AMP inside them so they release insulin - but only when blood glucose is already high, which is why it does not cause hypoglycaemia by itself. It also cuts the glucagon that pushes glucose up, and slows the stomach from emptying. That last effect, plus action on appetite signalling, is what produces both the weight loss and the nausea. (Source 1)

Is it good or bad for you?

Good, in the specific populations where it has been tested: in 9,340 people with type 2 diabetes at high cardiovascular risk it cut cardiovascular death from 6.0% to 4.7% over 3.8 years, and at the higher dose it produces 5 to 6 kg more weight loss than placebo over a year. Against that sit common gastrointestinal harms (nausea 39.3% versus 13.8% on placebo), a 37% higher pooled risk of gallbladder disease that more than doubles in weight-loss trials, a pancreatitis signal, and an unresolved rodent thyroid tumour finding. It is contraindicated outright in anyone with a personal or family history of medullary thyroid carcinoma or MEN 2. (Source 2)

How do you get more of it?

Only by prescription, as a once-daily injection under the skin of the abdomen, thigh or upper arm. There is no food, supplement or behaviour that provides liraglutide; your body makes its own GLP-1, but natural GLP-1(7-37) is broken down within minutes and makes up less than a fifth of circulating GLP-1. The dose is escalated in steps specifically to blunt nausea. (Source 1)

If it is harmful, what reduces it?

If it is causing harm, the response in the trials and the label is to stop the injections; there is no antidote and no taper. The label directs prescribers to stop it if the target dose cannot be tolerated, and to stop it in the weight indication if 4% of body weight has not been lost by 16 weeks. Stopping undoes the benefit as well: weight returns at about 0.4 kg a month. (Source 14)

Why might someone be low in it or missing it?

Liraglutide is a drug, not a nutrient, so no one is deficient in it. What can be low is the body's own GLP-1 response after meals, which is part of the physiology the drug is designed to replace - and the label notes that native GLP-1(7-37) is only a small fraction of circulating GLP-1 and is short-lived. In practice people stop having liraglutide because of nausea and vomiting: 9.8% discontinued for adverse reactions versus 4.3% on placebo. (Source 18)

Which whole foods contain it or feed it?

No food contains liraglutide. Food is relevant only in the other direction: the injection can be taken at any time of day independently of meals, and the drug slows the stomach, which is why portions feel larger and nausea is common. Every trial of its weight effect gave it alongside diet and activity changes - SCALE Diabetes used a 500 kcal per day deficit and at least 150 minutes a week of activity - so the trial results are the drug plus those changes, not the drug alone. (Source 20)

What happens if you do not have it?

There is no deficiency state. Not taking it means the benefits measured in the trials do not apply: in LEADER the placebo group had a 14.9% rate of cardiovascular death, heart attack or stroke against 13.0% on liraglutide, so the absolute difference over 3.8 years was 1.9 percentage points. For weight, people who stop regain about 0.4 kg a month and their cardiometabolic markers are projected back to baseline within about 1.4 years. (Source 4)

How can you test for it?

There is no blood test for a liraglutide level in routine care. What gets measured is the effect: HbA1c for the diabetes indication, body weight and BMI for the weight indication, with the label setting a 16-week weight check as the decision point. Calcitonin and thyroid ultrasound are sometimes suggested for the thyroid concern behind the boxed warning, but the label says explicitly that routine use of either is of uncertain value for picking up medullary thyroid cancer. (Source 2)

References

1. DailyMed / US FDA Structured Product Label, Novo Nordisk. VICTOZA (liraglutide) injection - full prescribing information including boxed warning (SPL version 31, published 17 November 2025) - section 12.1 Mechanism of Action. 2025. Read the source
2. DailyMed / US FDA Structured Product Label, Novo Nordisk. VICTOZA (liraglutide) injection - full prescribing information including boxed warning (SPL version 31, published 17 November 2025) - boxed warning. 2025. Read the source
3. Diabetes, Metabolic Syndrome and Obesity. A Systematic Review and Meta-Analysis of Efficacy and Safety of Liraglutide in Patients with Type 2 Diabetes Mellitus. 2026. PMID 41836360, DOI 10.2147/dmso.s570273. Read the source
4. The New England Journal of Medicine. Liraglutide and Cardiovascular Outcomes in Type 2 Diabetes. 2016. PMID 27295427, DOI 10.1056/nejmoa1603827. Read the source
5. The New England Journal of Medicine. A Randomized, Controlled Trial of 3.0 mg of Liraglutide in Weight Management. 2015. PMID 26132939, DOI 10.1056/nejmoa1411892. Read the source
6. The Lancet. Liraglutide safety and efficacy in patients with non-alcoholic steatohepatitis (LEAN): a multicentre, double-blind, randomised, placebo-controlled phase 2 study. 2016. PMID 26608256, DOI 10.1016/s0140-6736(15)00803-x. Read the source
7. JAMA. Effects of Liraglutide on Clinical Stability Among Patients With Advanced Heart Failure and Reduced Ejection Fraction: A Randomized Clinical Trial. 2016. PMID 27483064, DOI 10.1001/jama.2016.10260. Read the source
8. eClinicalMedicine. Effects of glucagon-like peptide-1 receptor agonists on alcohol consumption: a systematic review and meta-analysis. 2025. PMID 41324012, DOI 10.1016/j.eclinm.2025.103645. Read the source
9. DailyMed / US FDA Structured Product Label, Novo Nordisk. VICTOZA (liraglutide) injection - full prescribing information including boxed warning (SPL version 31, published 17 November 2025) - section 7 Drug Interactions. 2025. Read the source
10. DailyMed / US FDA Structured Product Label, Novo Nordisk. VICTOZA (liraglutide) injection - full prescribing information including boxed warning (SPL version 31, published 17 November 2025) - section 12.3 Drug Interaction Studies. 2025. Read the source
11. DailyMed / US FDA Structured Product Label, Novo Nordisk. VICTOZA (liraglutide) injection - full prescribing information including boxed warning (SPL version 31, published 17 November 2025) - section 12.3 Drug Interaction Studies (oral contraceptives). 2025. Read the source
12. DailyMed / US FDA Structured Product Label, Novo Nordisk. VICTOZA (liraglutide) injection - full prescribing information including boxed warning (SPL version 31, published 17 November 2025) - section 2 Dosage and Administration. 2025. Read the source
13. BMJ. Weight regain after cessation of medication for weight management: systematic review and meta-analysis. 2026. PMID 41500720, DOI 10.1136/bmj-2025-085304. Read the source
14. DailyMed / US FDA Structured Product Label, Novo Nordisk. SAXENDA (liraglutide) injection - full prescribing information including boxed warning (SPL version 22, published 15 June 2026) - section 2.1 Recommended Dosage. 2026. Read the source
15. JAMA Internal Medicine. Association of Glucagon-Like Peptide-1 Receptor Agonist Use With Risk of Gallbladder and Biliary Diseases: A Systematic Review and Meta-analysis of Randomized Clinical Trials. 2022. PMID 35344001, DOI 10.1001/jamainternmed.2022.0338. Read the source
16. Endocrinology, Diabetes & Metabolism. Evaluating the Rates of Pancreatitis and Pancreatic Cancer Among GLP-1 Receptor Agonists: A Systematic Review and Meta-Analysis of Randomised Controlled Trials. 2025. PMID 40988099, DOI 10.1002/edm2.70113. Read the source
17. Diabetes Care. GLP-1 Receptor Agonists and the Risk of Thyroid Cancer. 2023. PMID 36356111, DOI 10.2337/dc22-1148. Read the source
18. DailyMed / US FDA Structured Product Label, Novo Nordisk. SAXENDA (liraglutide) injection - full prescribing information including boxed warning (SPL version 22, published 15 June 2026) - section 6.1 Clinical Trials Experience. 2026. Read the source
19. DailyMed / US FDA Structured Product Label, Novo Nordisk. VICTOZA (liraglutide) injection - full prescribing information including boxed warning (SPL version 31, published 17 November 2025) - section 6.1 Clinical Trials Experience. 2025. Read the source
20. JAMA. Efficacy of Liraglutide for Weight Loss Among Patients With Type 2 Diabetes: The SCALE Diabetes Randomized Clinical Trial. 2015. PMID 26284720, DOI 10.1001/jama.2015.9676. Read the source
21. AACE Endocrinology and Diabetes. Incretin-Based Therapy and Thyroid Cancer Risk: A Systematic Review and Meta-Analysis of Randomized Controlled Trials. 2026. PMID 42221413, DOI 10.1016/j.aed.2026.03.003. Read the source
22. DailyMed / US FDA Structured Product Label, Novo Nordisk. VICTOZA (liraglutide) injection - full prescribing information including boxed warning (SPL version 31, published 17 November 2025) - section 2 Dosage and Administration. 2025. Read the source