Skip to content
GLP Loss
← Research
Evidence

GLP-1 vs Contrave: The Trial That Was Stopped Before It Could Answer

Naltrexone-bupropion's cardiovascular trial read as a 41% risk reduction at its first interim look and as nothing at its second. The sponsor leaked the first one, and the trial never reached the point where the question could be answered.

Owen Castellanos9 min read
One trial, two looks, and then no answerLIGHT: 8,910 adults, cardiovascular events on naltrexone-bupropionAt 25% of planned events, leaked by the sponsorHazard ratio 0.59 (95% CI, 0.39 to 0.90). 35 events vs 59.At 50% of planned events, after terminationHazard ratio 0.88 (99.7% CI, 0.57 to 1.34). 90 events vs 102.The question the trial was built to answerA margin of 1.4 at 378 events. Never reached. Never tested.The label still states the cardiovascular effect is not established.Semaglutide has a completed outcome trial. This drug has a stopped one.No randomized trial has compared the two against each other.

Contrave is two old generics in one extended-release tablet: naltrexone, an opioid receptor antagonist better known from alcohol use disorder treatment, and bupropion, an antidepressant. It is swallowed rather than injected, it is usually cheaper than a branded incretin, and it is the option a prescriber reaches for when a needle is refused. What it is not is a smaller version of the same thing. The effect size is roughly a third, the contraindication list is long enough to exclude a substantial share of the people asking, and the one trial that could have given it a cardiovascular claim was destroyed by its own sponsor.

What the COR trials measured, and who was still there at the end

COR-I randomized 1,742 adults to naltrexone 32 mg plus bupropion 360 mg daily, to a 16 mg naltrexone version, or to placebo for 56 weeks. Mean weight change was −6.1% on the 32 mg combination against −1.3% on placebo, with a 5% or greater loss reached by 48% against 16%.[1] COR-II reproduced it in 1,496 adults: −6.4% against −1.2% at week 56, with 50.5% reaching 5% against 17.1%.[2]

Now the number that rarely travels with those. Of the 1,742 people randomized in COR-I, 870 — exactly 50% — completed 56 weeks, and the completion rate was near-identical in all three arms.[1] The headline percentage is a modified-intention-to-treat analysis with last observation carried forward, computed over a trial that half its participants left. That is not a flaw unique to this drug, but it sets the scale for everything below it.

Two more COR trials fill in the range. COR-BMOD added intensive behavior modification to both arms and reported −9.3% against −5.1% on the modified-intention-to-treat analysis, −11.5% against −7.3% among completers, and −7.8% against −4.9% across all randomized participants — three different numbers for one trial, depending on who is counted.[3] COR-Diabetes, in 505 adults with type 2 diabetes, reported −5.0% against −1.8% and a glycated hemoglobin fall of 0.6 points against 0.1.[4]

The gap, stated without cushioning

A 2026 network meta-analysis of 262 randomized trials and 99,791 participants lists the agents that clear moderate-to-high certainty for one-year weight loss: tirzepatide at −14.9% (95% CI, −16.0 to −13.9), cagrilintide-semaglutide at −14.8%, oral semaglutide at −10.9%, orforglipron at −9.9%, subcutaneous semaglutide at −9.8% and phentermine-topiramate at −8.1%. Naltrexone-bupropion does not appear on that list.[5] It appears instead in the harms section, three times over: among the agents with the highest discontinuation for adverse events, among those with the most increased gastrointestinal events, and with the single largest fatigue signal of any drug analyzed — a relative risk of 8.9, an absolute increase of 331 per 1,000 people over one year.[5]

An earlier pooled analysis of 132 trials put naltrexone-bupropion’s odds of discontinuation for adverse events at 2.69 (95% CI, 2.10 to 3.44) against lifestyle modification alone — the highest of any drug class examined, above phentermine-topiramate at 2.40 and the GLP-1 receptor agonists at 2.22.[6] The comparison to set that against is in the semaglutide article.

The cardiovascular trial that answered nothing

LIGHT enrolled 8,910 overweight or obese adults at raised cardiovascular risk across 266 United States centers and randomized them to naltrexone-bupropion or placebo. Mean age was 61.0, 85.2% had diabetes, and 32.1% had established cardiovascular disease. The design planned a confidential interim look at roughly 87 events to assess a noninferiority hazard ratio of 2.0 for regulatory purposes, and a primary analysis at 378 events against a margin of 1.4.[7]

At the 25% interim, major adverse cardiovascular events had occurred in 59 placebo participants (1.3%) and 35 on naltrexone-bupropion (0.8%): hazard ratio 0.59 (95% CI, 0.39 to 0.90). The sponsor released that confidential interim result publicly. The trial’s academic leadership recommended termination and the sponsor agreed.[7]

At 50% of planned events the same trial read differently: 102 events (2.3%) on placebo against 90 (2.0%) on the drug, hazard ratio 0.88 (adjusted 99.7% CI, 0.57 to 1.34).[7] The apparent 41% reduction was an early-look artifact, and the trial never reached the event count at which its real question could be asked. The authors’ own conclusion is that noninferiority against the prespecified 1.4 margin cannot be assessed, that cardiovascular safety remains uncertain, and that a new adequately powered trial would be required.[7] Ten years later, none has reported, and the label still carries the limitation that the effect on cardiovascular morbidity and mortality has not been established.

That is the asymmetry that decides this comparison for anyone choosing on risk rather than on pounds. One drug in the field has a completed, published cardiovascular outcome trial with a positive primary endpoint, described in the SELECT article. The other has a trial that was stopped mid-course by a disclosure failure.

The contraindication list rules out more people than it looks like

The label opens with a boxed warning for suicidal thoughts and behaviors, carried over from the antidepressant class that bupropion belongs to, and states that Contrave is not approved for use in pediatric patients. Contraindications then run: uncontrolled hypertension; a seizure disorder or any history of seizures; use of any other bupropion-containing product; bulimia or anorexia nervosa; chronic opioid or opiate agonist or partial agonist use, or acute opiate withdrawal; abrupt discontinuation of alcohol, benzodiazepines, barbiturates or antiepileptic drugs; monoamine oxidase inhibitors within 14 days; and known allergy to either component.

Read the list against who actually buys weight medication online. Uncontrolled hypertension is common in this population. A history of seizures is absolute, not relative. Bulimia and anorexia nervosa are prior diagnoses rather than current ones, so a binge-eating history recorded years ago still closes the door. Chronic opioid use rules out anyone on maintenance buprenorphine or methadone, and an opioid antagonist will precipitate withdrawal in anyone who did not disclose it. None of these applies to the incretin class, whose own exclusions are a different and shorter list set out in the contraindications article.

Two further label facts belong with the dosing. Seizure incidence in the Contrave trials was approximately 0.1% against 0% on placebo, which is why the four-week escalation exists, why no more than two tablets may be taken at once, and why the label warns against taking it with a high-fat meal — the food effect raises systemic exposure to both components. And blood pressure: against placebo, systolic differences ran +1.8 to +2.4 mmHg and diastolic +1.7 to +2.1 mmHg during the first twelve weeks (all p < 0.001), with heart rate 2.1 beats per minute higher at weeks 4 and 8 and 1.7 higher at week 52.

Two instruments on suicidality, pointing opposite ways

A 2026 disproportionality analysis of roughly 78,000 anti-obesity reports in the FDA Adverse Event Reporting System found naltrexone-bupropion with a reporting odds ratio of 3.84 (95% CI, 2.89 to 5.12) for suicidal ideation and 4.11 (95% CI, 1.62 to 10.45) for suicide attempt. Semaglutide’s ideation ratio was 1.39 (95% CI, 0.99 to 1.94) and not significant; tirzepatide’s ratios were below one for every outcome.[8]

The label’s own pooled trial data say something different. Across placebo-controlled obesity trials of up to 56 weeks, no suicides or suicide attempts were reported, and suicidal ideation was reported by 3 of 1,515 on placebo (0.20%) against 1 of 3,239 on Contrave (0.03%). Both statements are accurate. A reporting odds ratio measures how often an event is written down for a drug relative to other drugs, not how often it happens; a drug carrying a boxed warning about a harm attracts reports of that harm. A randomized trial measures incidence but is too small and too short to detect a rare event. Neither instrument answers the other’s question, and how this signal has been handled across the class is set out in the mental health article.

Where the guidelines put it

The American College of Physicians living guideline of April 2026 ranks, for adults with obesity, semaglutide and tirzepatide first-line on moderate-certainty evidence, then phentermine-topiramate, then liraglutide, and naltrexone-bupropion fourth-line on low-certainty evidence — last of the five named agents. For adults with overweight at a body-mass index of 27 to under 30 plus a weight-related condition, naltrexone-bupropion does not appear at any line. The guideline names suicidal ideation with naltrexone-bupropion among the specific warnings a clinician should raise before prescribing.[9]

Nobody has run the comparison

No randomized trial has assigned adults to naltrexone-bupropion or to a GLP-1 receptor agonist. The living systematic review behind that guideline covered 69 studies and 112,511 participants and states its own limitation: direct head-to-head comparisons of different treatments were limited.[10] Every figure above that sets the two side by side is an indirect estimate from separate trials in separate populations.

Real-world persistence is where the difference shows up without needing a trial. A retrospective cohort of 1,911 adults starting an anti-obesity medication found overall persistence of 44% at three months, 33% at six and 19% at one year. Semaglutide had the highest 1-year persistence at 40%; against phentermine-topiramate as the reference, semaglutide’s adjusted odds of 1-year persistence were 4.26 (95% CI, 3.04 to 6.05) and naltrexone-bupropion’s were 0.68 (95% CI, 0.46 to 1.00).[11] A separate claims and records analysis of 1,563 new anti-obesity prescriptions found that 91.1% were never filled at all within 60 days, with naltrexone-bupropion the most-prescribed agent in the cohort at 36.3% — and no significant difference in non-filling between drugs (p = 0.299).[12]

Who each one suits

Contrave is an oral, non-injected option with a distinctive mechanism aimed at reward-driven eating rather than at satiety hormones, it is available to people who will not use a needle, and in cost-effectiveness models it sits among the cheapest agents. Against that, its trial effect is roughly a third of what the top incretin doses produce, its discontinuation profile is the worst in the class, its contraindication list excludes a substantial group, and it has no cardiovascular outcome evidence because the trial that would have produced it was terminated.

Every incretin figure quoted above came from a trial of an FDA-approved product at a labeled dose. Compounded semaglutide and tirzepatide are not FDA-approved and are not reviewed by the FDA for safety, efficacy or quality before dispensing, so a compounded vial priced near a Contrave prescription does not carry the trial evidence with it. How the figures on this site are established before publication is described in the methodology.

Frequently asked

How much weight does Contrave take off compared with a GLP-1?
Roughly a third as much at the top doses. COR-I reported a mean change of −6.1% at 56 weeks against −1.3% on placebo, and COR-II −6.4% against −1.2%. A 2026 network meta-analysis of 262 trials put tirzepatide at −14.9% against lifestyle alone and did not list naltrexone-bupropion among the agents with moderate-to-high certainty for weight loss at one year.
Did the LIGHT trial show Contrave protects the heart?
No. At the 25% interim analysis the hazard ratio for major adverse cardiovascular events was 0.59 (95% CI, 0.39 to 0.90), and that confidential result was released publicly by the sponsor, which led to the trial's termination. At 50% of planned events the hazard ratio was 0.88 (99.7% CI, 0.57 to 1.34), and the prespecified noninferiority margin of 1.4 was never testable. The label still states the cardiovascular effect has not been established.
Who cannot take Contrave?
The label contraindicates it in uncontrolled hypertension, any seizure disorder or history of seizures, use of another bupropion product, bulimia or anorexia nervosa, chronic opioid or opiate agonist or partial agonist use or acute opiate withdrawal, abrupt discontinuation of alcohol, benzodiazepines, barbiturates or antiepileptic drugs, monoamine oxidase inhibitor use within 14 days, and known allergy to either component. Seizure incidence in the trials was approximately 0.1% against 0% on placebo.
Does Contrave cause suicidal thoughts?
The two available instruments disagree. A 2026 analysis of the FDA Adverse Event Reporting System found a reporting odds ratio of 3.84 (95% CI, 2.89 to 5.12) for suicidal ideation on naltrexone-bupropion, while the label's own pooled trials recorded ideation in 1 of 3,239 people on Contrave against 3 of 1,515 on placebo, with no suicides or attempts. A reporting ratio measures how often an event is written down, not how often it occurs, and the drug carries a boxed warning that itself prompts reporting.
Do people stay on Contrave?
Mostly not. In a cohort of 1,911 adults starting an anti-obesity medication, overall persistence was 44% at three months and 19% at one year, with naltrexone-bupropion at adjusted odds of 0.68 (95% CI, 0.46 to 1.00) for 1-year persistence against phentermine-topiramate, while semaglutide was at 4.26. A separate analysis of 1,563 new prescriptions found 91.1% were never filled within 60 days at all.
Has Contrave ever been compared to semaglutide in a trial?
No randomized trial has assigned adults to naltrexone-bupropion or to any GLP-1 receptor agonist. The living systematic review supporting the 2026 American College of Physicians guideline covered 69 studies and 112,511 participants and states that direct head-to-head comparisons between treatments were limited. Every side-by-side figure is an indirect estimate pooled across separate trials.

Sources

  1. [1] Greenway FL, Fujioka K, Plodkowski RA, et al. (2010). Effect of naltrexone plus bupropion on weight loss in overweight and obese adults (COR-I): a multicentre, randomised, double-blind, placebo-controlled, phase 3 trial. Lancet. PMID 20673995
  2. [2] Apovian CM, Aronne L, Rubino D, et al. (2013). A randomized, phase 3 trial of naltrexone SR/bupropion SR on weight and obesity-related risk factors (COR-II). Obesity (Silver Spring). PMID 23408728
  3. [3] Wadden TA, Foreyt JP, Foster GD, et al. (2011). Weight loss with naltrexone SR/bupropion SR combination therapy as an adjunct to behavior modification: the COR-BMOD trial. Obesity (Silver Spring). PMID 20559296
  4. [4] Hollander P, Gupta AK, Plodkowski R, et al. (2013). Effects of naltrexone sustained-release/bupropion sustained-release combination therapy on body weight and glycemic parameters in overweight and obese patients with type 2 diabetes. Diabetes Care. PMID 24144653
  5. [5] Nong K, Shi Q, Zeng L, et al. (2026). Comparative effects of drugs for adults with overweight or obesity: systematic review and network meta-analysis. BMJ. PMID 42419792
  6. [6] Shi Q, Wang Y, Hao Q, et al. (2024). Pharmacotherapy for adults with overweight and obesity: a systematic review and network meta-analysis of randomised controlled trials. Lancet. PMID 38582569
  7. [7] Nissen SE, Wolski KE, Prcela L, et al. (2016). Effect of Naltrexone-Bupropion on Major Adverse Cardiovascular Events in Overweight and Obese Patients With Cardiovascular Risk Factors: A Randomized Clinical Trial. JAMA. PMID 26954408
  8. [8] Seijas-Amigo J, Rodríguez-Vázquez A, Cordero A, et al. (2026). Comparative pharmacovigilance analysis of suicidality-related adverse events among GLP-1 and non-GLP-1 anti-obesity drugs in the FDA Adverse Event Reporting System. Int J Clin Pharm. PMID 41739406
  9. [9] Qaseem A, Cross JT Jr, Harrod CS, et al. (2026). Pharmacologic Treatments With Lifestyle Modifications in Nonpregnant Adults With Overweight or Obesity in Outpatient Settings: A Living Clinical Guideline From the American College of Physicians (April 2026). Ann Intern Med. PMID 42296496
  10. [10] Damen JAA, Hooft L, Jenniskens K, et al. (2026). Benefits and Harms of Pharmacologic Treatments in Adults With Overweight or Obesity: A Living Systematic Review and Network Meta-analysis for the American College of Physicians. Ann Intern Med. PMID 42296503
  11. [11] Gasoyan H, Pfoh ER, Schulte R, et al. (2024). Early- and later-stage persistence with antiobesity medications: A retrospective cohort study. Obesity (Silver Spring). PMID 38053443
  12. [12] Kan H, Bae JP, Dunn JP, et al. (2023). Real-world primary nonadherence to antiobesity medications. J Manag Care Spec Pharm. PMID 37594848

Where to get it

Best GLP-1 injections

Every injectable seller we can verify, with the price each one publishes and an honest read of what the trials measured.

Compare providers →

More in Evidence