Retatrutide Knee Osteoarthritis Trials in NEJM: What the TRIUMPH-4 and TRIUMPH-1-OA Paper Reports

What Did NEJM Publish on 8 November 2026?

The New England Journal of Medicine published “Retatrutide Once Weekly in Adults with Knee Osteoarthritis Pain and Obesity” (DOI 10.1056/NEJMoa2610676), timed to Eli Lilly’s plenary presentation at the American College of Rheumatology’s ACR Convergence 2026 meeting.[1][6] The paper reports two sponsor-funded Phase 3 trials of retatrutide (LY3437943), Lilly’s investigational GIP, GLP-1 and glucagon receptor agonist, in adults with knee osteoarthritis (OA) and obesity:

TRIUMPH-4 (NCT05931367): 445 participants randomised to once-weekly retatrutide 9 mg, 12 mg or placebo, with the primary analysis at week 68.[1][3]

TRIUMPH-1-OA: 574 participants with knee OA enrolled within the broader TRIUMPH-1 obesity trial (NCT05929066), randomised to 4 mg, 9 mg, 12 mg or placebo, with the WOMAC pain endpoint assessed at week 80.[1][4]

Our ACR Convergence 2026 preview, written before the abstracts were public, set out what to check once full data appeared. This guide follows up with what the peer-reviewed abstract actually reports.

What Were the Reported WOMAC Pain Results?

WOMAC pain is a patient-reported subscale of the Western Ontario and McMaster Universities Osteoarthritis Index, expressed here on a 0–10 scale. The figures below are as reported in the NEJM abstract and the ACR meeting abstract.[1][2]

Body weight was a co-primary endpoint in TRIUMPH-4. Weight data are covered in our existing TRIUMPH-1 NEJM guide and are not the focus here.

How Does the Paper Differ From the December 2025 Topline?

Lilly’s 11 December 2025 TRIUMPH-4 press release led with a percentage figure: a 75.8% reduction in WOMAC pain from baseline with the 12 mg dose.[5] The NEJM abstract leads with absolute point changes and a placebo comparator. Both can be accurate descriptions of the same dataset, but they answer different questions.

The absolute figures make the placebo arm visible. In TRIUMPH-4, the placebo group’s mean WOMAC pain score fell by 2.12 points, and in the pooled responder analysis 37.8% of placebo participants reached a 50% or greater pain reduction.[1] That is a large placebo response, though not unusual for patient-reported OA pain endpoints. For researchers, the between-group difference (roughly 1.6–1.8 points in TRIUMPH-4) is the quantity that the trial’s randomisation actually supports. Our original TRIUMPH-4 knee OA guide covers the topline release in detail.

Why Do the TRIUMPH-1-OA Numbers Differ Between Sources?

The NEJM abstract and the ACR meeting abstract give different week-80 values for the same sub-study: −3.64 versus −4.30 points for 12 mg, and −1.87 versus −2.24 for placebo.[1][2] The NEJM figures are labelled as a hybrid estimand. In incretin trials, Lilly typically reports a treatment-regimen-style estimand (all randomised participants, regardless of whether they stayed on treatment) and an efficacy-style estimand (as if participants had remained on treatment). The ACR values are larger in every arm, which is consistent with an on-treatment analysis, but we could not read the full methods to confirm this.

The practical point: when citing these trials, state the source and the estimand alongside the number. Mixing figures from the two abstracts would overstate or understate the effect depending on direction. The same issue arises in cross-trial comparisons, covered in our cross-trial limits guide.

What Are the Limits of the Pooled Analysis?

The headline responder figure (63.2%–69.6% versus 37.8%) is described as pooled at week 68, yet TRIUMPH-1-OA’s primary timepoint is week 80.[1] Pooling a dedicated OA trial with a sub-study embedded in an obesity trial combines populations recruited under different protocols, with different dose sets (TRIUMPH-4 had no 4 mg arm). Our basket trial design explainer sets out why a sub-study within a trial powered for another endpoint supports narrower inferences.

Completion also matters. Only 72% of TRIUMPH-4 and 75% of TRIUMPH-1-OA participants completed assigned treatment.[1] With roughly a quarter of participants discontinuing, the handling of missing data drives much of the gap between estimands. December 2025 topline reporting gave adverse-event discontinuation in TRIUMPH-4 of 12.2% (9 mg) and 18.2% (12 mg) versus 4.0% with placebo.[5] Differential discontinuation and gastrointestinal effects can also weaken blinding on a self-reported outcome.

What Do the Trials Not Show?

Trade-press coverage of the publication noted that neither trial was designed to test whether retatrutide slows structural knee OA progression or reduces the need for total joint replacement, and that persistence of any change after treatment stops is unknown.[7] Because retatrutide also lowers body weight, these trials cannot separate a weight-mediated pain change from any direct joint or neural mechanism; any mediation analysis would be exploratory. Preclinical work on incretin receptor signalling in joint tissue remains the more appropriate setting for mechanistic questions, and our mechanism of action guide summarises the receptor pharmacology.

What Is Retatrutide’s Regulatory Status?

Retatrutide has no marketing authorisation anywhere. In its 23 July 2026 release, Lilly described a planned Biologics License Application to the FDA in the first quarter of 2027.[8] A journal publication is not a regulatory decision. Our FDA approval timeline guide tracks the submission pathway.

How Does Licensed Pharma Development Differ From Research-Use Compounds?

TRIUMPH-4 and TRIUMPH-1 are regulated clinical development by a licensed pharmaceutical sponsor, with registered protocols, ethics oversight and medical supervision of participants. Their results do not transfer to a laboratory reagent. Velox Peptides supplies retatrutide solely as an HPLC-verified in vitro research reagent and reports trial news only as third-party information for researchers. See the research-grade retatrutide page for specifications.

This compound is supplied as a research reagent only. It is not a medicine and has not been evaluated by the MHRA or FDA. Not for human or veterinary use. See our Research Use Policy and MHRA Statement.

Frequently Asked Questions