Retatrutide eased knee pain in two big trials. Was it just the weight loss?

By Peptide Policy Watch staff · Published November 9, 2026 · 7 min read · 8 sources, 8 checked at the source

An empty running track at dawn, with mist, a low sun behind trees, empty grandstands and a pair of metal starting blocks on the red lanes.

Bad knees and extra weight feed each other. Every stair hurts, so you move less, so the weight stays. That's why rheumatologists paid attention on November 8, 2026, when two phase 3 trials of retatrutide, Lilly's experimental triple-hormone drug, landed in the New England Journal of Medicine and on the main stage of the American College of Rheumatology meeting in Orlando. Pain scores dropped a lot. So did body weight. The interesting question is how much of the first came from the second.

Facts are as of November 9, 2026. Sources, with retrieval dates, are at the bottom.

If you only read one box

Two randomized, placebo-controlled phase 3 trials, TRIUMPH-4 (445 people, 68 weeks) and TRIUMPH-1-OA (574 people, 80 weeks), enrolled adults with painful knee osteoarthritis and obesity or overweight, and no diabetes. [V]
On a 0-to-10 knee pain scale where people started near 6, pain fell by roughly 3.2 to 4 points on retatrutide versus about 2 on placebo. Weight fell about 17% to 25% versus about 4% to 5%. [V] (Large RCTs.)
Serious side effects and treatment stops were more common on retatrutide than placebo. Stomach and gut effects led the list. [V]
Retatrutide is not FDA-approved for anything. Lilly says it plans to file in the first quarter of 2027, and FDA says it can't be compounded. [V]

The numbers first

Both trials used the WOMAC pain score, a standard knee osteoarthritis questionnaire scored here from 0 to 10, where higher means more pain. To get in, people needed moderate to severe knee pain (4 to 9), X-ray signs of osteoarthritis, a BMI of at least 27 plus a weight-related health problem, and at least one failed attempt at dieting. About four in five had a BMI above 35. [V]

	TRIUMPH-4	TRIUMPH-1-OA
People randomized	445	574
Length	68 weeks	80 weeks
Retatrutide groups	Two strengths	Three strengths
Knee pain change, retatrutide	−3.74 to −3.96 points	−3.22 to −3.64 points
Knee pain change, placebo	−2.12 points	−1.87 points
Weight change, retatrutide	−20.0% to −23.7%	−17.5% to −24.8%
Weight change, placebo	−4.6%	−4.3%

Figures above use the trials' main analysis methods, as reported in the authors' accepted manuscript. Every retatrutide group beat placebo on pain and weight. [V] (Large RCTs; Lilly funded both, and several authors work for Lilly.) [V]

The pooled result is the one most people will quote. At week 68, between 63.2% and 69.6% of people on the two higher strengths had cut their knee pain at least in half, compared with 37.8% on placebo. Physical function, which covers things like stairs, standing up and walking, followed the same pattern: 63.5% to 64.6% versus 33.1%. [V]

69.6%of one retatrutide group cut knee pain at least in half by week 68 (pooled), vs 37.8% on placebo [V]
1,019adults across the two trials [V]
0FDA approvals for retatrutide so far [V]

Lilly's own release adds a striking line: more than one in eight retatrutide patients in TRIUMPH-4 reported no knee pain at all by the end. Lilly notes that its percentage figures came from an analysis that wasn't planned in advance. [V]

So was it the weight?

Here's where the paper gets honest. The authors write that how GLP-1-type drugs reduce pain is incompletely understood. [V] Losing weight takes load off the knee, and the load drop is several times the pounds lost; the paper cites earlier work putting it near fourfold per pound. [V] There's also a second idea: shrinking fat tissue may lower body-wide inflammation, and some data hint that GLP-1 drugs have anti-inflammatory effects of their own. The authors call for more study to sort out which of these matter. [V]

What the trials can't do is separate the two. Nobody in them lost a quarter of their weight without the drug, so we can't tell whether retatrutide does anything for knees beyond the scale. [U] That's our reading of the design, not a claim from the paper.

For context, the authors compared their effect sizes with older trials: a statistical effect size of 0.55 to 0.79 here, against 0.37 for maximum-strength ibuprofen, 0.31 for tramadol, 0.61 for steroid shots into the joint and 0.50 for semaglutide in its own knee trial. They also wrote that direct comparisons aren't feasible, because the studies differ. [V] Different trials, different people, different methods; treat that list as background, not a ranking.

The placebo groups improved too, by about two points. That's common in pain trials, and it's a reminder that the gap between the arms, not the raw drop, is the drug's measured effect. [U] (General point about trial design.)