For knee arthritis with obesity, the trials point mostly one way: people on semaglutide or retatrutide reported less knee pain than people on placebo. But one trial found no pain benefit at all, and outside the knee there is almost nothing. No GLP-1 medicine is approved for arthritis. The evidence is three knee programmes, each in people with obesity, each unable to say how much of the benefit is simply weight. For hip arthritis, gout and rheumatoid arthritis, a registry search on 2 November 2026 found no registered GLP-1 trial.
This page reports the trials as published or announced, each dated. It does not suggest anyone start a GLP-1 for joint pain; that is a decision for a clinician. Whether these drugs can cause aches is a different question, covered on GLP-1s and joint pain.
The three knee programmes
| Trial | Who | Length | Drug and dose arm | Weight change, drug vs placebo | Knee pain change, drug vs placebo | Pain scale |
|---|---|---|---|---|---|---|
| STEP 9 (Bliddal 2024, NEJM) | 407 adults, BMI ≥30, moderate knee OA with at least moderate pain | 68 weeks | semaglutide 2.4 mg weekly | −13.7% vs −3.2% | −41.7 vs −27.5 points | WOMAC pain, 0–100 |
| Liraglutide after diet (Gudbergsen 2021, AJCN) | 156 adults, BMI ≥27, knee OA, who first lost >5% on an 8-week diet | 52 weeks | liraglutide 3 mg daily | −2.8 kg vs +1.2 kg | +0.4 vs −0.6 points (no difference) | KOOS pain |
| TRIUMPH-4 (Lilly release, 11 Dec 2025) | 445 adults, overweight or obesity, knee OA | 68 weeks | retatrutide 9 mg / 12 mg weekly | −26.4% / −28.7% vs −2.1% (efficacy estimand) | −4.5 / −4.4 vs −2.4 points | WOMAC pain, 0–10 |
| TRIUMPH-1, knee OA subgroup (NEJM 2026) | 574 of 2,339 adults with obesity, no diabetes | 80 weeks | retatrutide 9 mg / 12 mg | whole trial: −23.7% / −25.0% vs −3.9% (ITT) | −3.9 / −4.1 vs −2.5 points (ITT) | WOMAC pain, 0–10 |
Dose figures are the trial arms, not recommendations. Sources: STEP 9, PMID 39476339; liraglutide trial, PMID 33471039; TRIUMPH-4 figures from Lilly's topline release as reported by Healio (Lilly's own release page refused an automated read on 2 November 2026; the same figures are on this site's retatrutide page); TRIUMPH-1, PMID 42814954.
The pain scales are not the same size. STEP 9 scored WOMAC pain from 0 to 100; the retatrutide trials use a 0–10 version. A fall of 4 points on the second is not smaller than a fall of 40 on the first. Within each trial, the gap between drug and placebo is the number to read.
STEP 9: the trial built for it
STEP 9 was run at 61 sites in 11 countries and funded by Novo Nordisk. Participants were mostly women (81.6%), mean age 56, mean BMI 40.3, and started with a mean WOMAC pain score of 70.9 out of 100: painful knees, not mild ones. Both groups had diet and activity counselling.
- Weight: −13.7% against −3.2% at 68 weeks.
- Knee pain: −41.7 against −27.5 points.
- Physical function (SF-36): +12.0 against +6.5 points.
- Stopped for adverse events: 6.7% against 3.0%, mostly gastrointestinal.
Two limits sit inside those numbers. The placebo group's pain fell by 27.5 points on its own, so most of the improvement came with taking part in a trial with counselling. And the trial cannot tell weight loss apart from any other effect of the drug: it was not designed to.
The trial that found nothing on pain
The Danish liraglutide trial (registered as NCT02905864) is the one most summaries leave out. It first put 168 people on an eight-week low-calorie diet; both knee pain and weight fell during that phase. The 156 who lost more than 5% were then randomised to liraglutide 3 mg a day or placebo for a year.
Liraglutide did what it was given for: weight changed by −2.8 kg against +1.2 kg, a gap of 3.9 kg. Knee pain did not differ (KOOS pain +0.4 against −0.6 points; difference 0.9, 95% CI −3.9 to 5.7). Gastrointestinal events were 50.2% against 39.2%.
The design differs from STEP 9: pain had already fallen with the diet before anyone took the drug, and the drug was measured on keeping weight off rather than taking it off. That is the honest reading, and it is also why this trial matters. It shows that a GLP-1 can move weight without moving knee pain, which is the question STEP 9 could not answer.
Retatrutide: bigger numbers, earlier stage
Retatrutide is investigational and has no FDA approval. Its knee results come from two sources at different stages:
- TRIUMPH-4 (445 people, 68 weeks) is a topline company release from 11 December 2025. It reports −4.5 and −4.4 WOMAC points on 9 and 12 mg against −2.4 on placebo (efficacy estimand) and −4.0 and −3.7 against −2.1 counting everyone regardless of stopping; 14.1% and 12.0% were pain-free at 68 weeks against 4.2% in a post hoc analysis. Stopping for adverse events was 12.2% and 18.2% against 4.0%. ClinicalTrials.gov (NCT05931367) lists primary completion on 14 November 2025 and no posted results on 2 November 2026.
- TRIUMPH-1 is peer-reviewed (NEJM, 29 September 2026). Its knee figures are for a subgroup of 574 with knee osteoarthritis inside a larger obesity trial; by intention to treat, WOMAC pain changed −3.9 and −4.1 on 9 and 12 mg against −2.5 on placebo.
Retatrutide's side-effect record, by dose arm, is on retatrutide side effects. The discontinuation rate at 12 mg in TRIUMPH-4 is the highest in any knee trial on this page.
What has not been tested
A ClinicalTrials.gov search on 2 November 2026 (conditions as below, interventions semaglutide, tirzepatide, liraglutide, dulaglutide or retatrutide):
| Condition | Registered trials found | Note |
|---|---|---|
| Osteoarthritis (any joint) | 17 records | every record that names a joint names the knee: STEP 9, TRIUMPH-4, STOP KNEE-OA, two trials of 4P004, a product injected into the knee joint, and imaging and biomarker sub-studies of the liraglutide trial; TRIUMPH-1 appears because its knee subgroup is registered |
| Hip osteoarthritis | 0 | |
| Gout | 0 | |
| Rheumatoid arthritis | 0 | |
| Psoriatic arthritis | 6 | one withdrawn; the largest, TOGETHER-PsA (tirzepatide with ixekizumab, 279 people), completed in November 2025 with no results posted |
Tirzepatide has no knee result yet. STOP KNEE-OA (NCT06191848, 352 participants, University of Melbourne) lists its primary completion as May 2027. A 54-person trial of a GLP-1 before knee replacement (NCT06807970) lists July 2027.
No trial on this page measured cartilage or joint structure as its main outcome. So whether GLP-1 drugs change the course of osteoarthritis, rather than how painful it feels, is not known. Laboratory work on inflammation is reviewed in a 2025 paper in Pharmacotherapy (PMID 39980227); a 2026 review of GLP-1s and chronic pain concluded that they "are not established analgesics" (PMID 42590035). Inflammation markers in the main GLP-1 trials are a separate subject.
What this means, and what it does not
The record supports one narrow sentence: in people with obesity and painful knee osteoarthritis, semaglutide and retatrutide reduced knee pain more than placebo over about 68 to 80 weeks. It does not support a claim about other joints, about people without obesity, about disease progression, or that pain relief is independent of weight. Arthritis treatment, and whether a weight medicine has a place in it, is a decision for a clinician who knows the person's joints and health record.
Related on this site: Wegovy weight-loss results (the STEP programme), is a GLP-1 safe?, and GLP-1 benefits.
Sources
- Bliddal H et al. Once-weekly semaglutide in persons with obesity and knee osteoarthritis (STEP 9). N Engl J Med 2024;391:1573–1583. PMID 39476339; NCT05064735.
- Gudbergsen H et al. Liraglutide after diet-induced weight loss for pain and weight control in knee osteoarthritis: a randomized controlled trial. Am J Clin Nutr 2021;113:314–323. PMID 33471039; NCT02905864.
- Retatrutide TRIUMPH-1. N Engl J Med, 29 September 2026. PMID 42814954.
- Lilly topline release on TRIUMPH-4, 11 December 2025, as reported by Healio; NCT05931367.
- Ryan M et al. The potential role of GLP-1 receptor agonists in osteoarthritis. Pharmacotherapy 2025;45:177–186. PMID 39980227.
- Hyung SW et al. GLP-1 receptor agonists and chronic pain: a narrative review. J Clin Med 2026;15:5932. PMID 42590035.
- ClinicalTrials.gov API v2, searched 2 November 2026.