Peptides for Arthritis Research

Peptides have become one of the busiest corners of arthritis research, and one of the most over-sold. Some have genuine randomised human trial data behind them. Some have a large animal literature and almost nothing in people. At least one reached phase 3 and failed. And one of the most useful peptides in rheumatology isn't a treatment at all — it's the basis of a blood test. This guide sorts the compounds by the strength of the evidence rather than the strength of the marketing, so you can see exactly where each one stands.

Two diseases, two different research targets

“Arthritis” covers very different conditions, and that matters before any peptide enters the conversation. According to the Australian Bureau of Statistics, just under 3.7 million Australians (14.5%) had arthritis in 2022. Osteoarthritis was the most common type of arthritis (57.0%), with rheumatoid arthritis at 13.9%.

Osteoarthritis is a degenerative joint disease in which cartilage breaks down and the joint structure changes; weight is a major factor, and losing weight has been shown to ease its symptoms, including pain. Rheumatoid arthritis is autoimmune: a chronic inflammatory disease of the joints that leads to destruction of cartilage and bone. A compound that calms an inflammatory cascade in a rat model may be relevant to one and nearly irrelevant to the other, so the first question to ask of any study is which disease the model actually represents.

The peptide already central to arthritis medicine

Before looking at therapies, it's worth knowing that peptides are already embedded in how rheumatoid arthritis is diagnosed. The anti-CCP test measures autoantibodies against cyclic citrullinated peptide (CCP), and it has become one of the key markers for diagnosing rheumatoid arthritis.

A meta-analysis of 37 studies in the Annals of Internal Medicine found anti-CCP antibodies had a pooled sensitivity of 67% — roughly two in three people with RA test positive — and a specificity of 95%, against 69% and 85% for IgM rheumatoid factor. In plain terms, a positive anti-CCP result is far less likely to be a false alarm, and three of four studies found anti-CCP positivity predicted radiographic progression better than rheumatoid factor. Serologic abnormality is one of the four scored domains in the 2010 ACR/EULAR classification criteria, worth up to 3 of the 10 points.

The strongest human evidence: semaglutide in knee osteoarthritis

Of the peptides in this guide, the strongest positive arthritis trial belongs to a GLP-1 receptor agonist. Semaglutide is a peptide analogue of the hormone GLP-1, and the STEP 9 trial, published in the New England Journal of Medicine, tested it directly in knee osteoarthritis.

STEP 9 was a 68-week, double-blind, placebo-controlled trial across 61 sites in 11 countries. It enrolled 407 people with obesity (mean BMI 40.3) and moderate knee osteoarthritis with at least moderate pain; 81.6% were women. Participants received once-weekly semaglutide 2.4 mg or placebo, and both groups got counselling on diet and physical activity. The results:

• Body weight: −13.7% with semaglutide versus −3.2% with placebo.
• WOMAC pain score (0–100): −41.7 points versus −27.5 points.
• SF-36 physical function: +12.0 points versus +6.5 points.
• Stopping treatment due to adverse events: 6.7% versus 3.0%, mostly gastrointestinal.

Two details are easy to miss. The placebo group's pain still fell by 27.5 points, a reminder of how much diet, activity counselling and trial participation do on their own — and why uncontrolled studies in this field are so hard to interpret. And because weight loss was one of the trial's two primary end points, it wasn't designed to separate how much of the pain benefit came from losing weight versus any other effect of the drug. The trial was funded by Novo Nordisk.

Collagen peptides: a modest effect, and not all collagen is the same

Oral collagen is the most accessible peptide product in this space, and it has a real, if modest, randomised evidence base. A meta-analysis of randomised placebo-controlled trials found collagen reduced the total WOMAC score (weighted mean difference −8.00; 95% CI −13.04 to −2.95) and visual analogue pain scores (−16.57; 95% CI −26.24 to −6.89). Look at the subscores, though, and the picture is more specific: stiffness improved significantly, but the WOMAC pain subscore (p = 0.75) and physical function subscore (p = 0.81) did not.

“Collagen” also covers two quite different things. Hydrolysed collagen is broken down into small peptides, whereas undenatured type II collagen (UC-II) is an intact form derived from chicken sternum cartilage. In a multicentre randomised trial of 191 people with knee osteoarthritis, 40 mg of UC-II daily for 180 days reduced the total WOMAC score significantly more than placebo (p = 0.002) and more than glucosamine plus chondroitin (p = 0.04), with improvements across the pain, stiffness and function subscales. The investigators noted that its mechanism still needs to be worked out. Dose and form differ so much between products that a result for one says little about another.

BPC-157: a large animal literature and one small human chart review

BPC-157, a 15-amino-acid fragment of a gastric protein, has the deepest musculoskeletal research record of the research-stage peptides. The key studies below all come from one research group in Zagreb, and all are in rats. Individually, they are striking. After complete transection of the Achilles tendon, rats given BPC-157 showed better load to failure, better function scores and fuller restoration of tendon integrity than controls. When the Achilles tendon was detached from the heel bone — an injury that does not heal spontaneously in that model — BPC-157 improved tendon-to-bone healing and reduced the healing damage caused by methylprednisolone.

The most arthritis-relevant study used adjuvant arthritis in rats, an inflammatory model. BPC-157 reduced lesion development after a single dose and more so with daily dosing, and when given as treatment for arthritis that was already established, the benefit appeared after two weeks and was still visible after a year of administration.

The human picture is very different. A 2025 systematic review in HSS Journal identified 36 eligible studies on BPC-157 in orthopaedic sports medicine: 35 preclinical and one clinical. It reported a half-life of under 30 minutes, found no clinical safety data, and noted that its use is banned in professional sport. The single clinical study is a retrospective chart review from a US clinic: of 17 patients who had received intra-articular injections, 16 were reached by phone and asked to recall their pain. Of the 12 given BPC-157 alone, 11 reported significant improvement. There was no control group, no validated outcome measure and a mix of knee conditions — so it's a reason to run a proper trial, not evidence of what that trial would find.

Thymosin beta-4 (TB-500): the joint data point somewhere unexpected

Thymosin beta-4 promotes angiogenesis and tissue repair in preclinical models, and it's commonly sold in blends with BPC-157. What rarely gets mentioned is what happens to thymosin beta-4 inside arthritic joints: it goes up.

In 216 people with knee osteoarthritis and 152 healthy controls, serum thymosin beta-4 was higher in the osteoarthritis group, and levels in both serum and joint fluid rose with radiographic severity. In synovial fluid from rheumatoid arthritis, mean levels were 1,359 ng/mL against 145 ng/mL in osteoarthritis (with very wide variation between patients), and in RA — but not osteoarthritis — they tracked with MMP-9, MMP-13, VEGF, IL-6 and IL-8 — molecules involved in inflammation and bone erosion. Correlation isn't causation; the rise may well be part of the joint's attempt at repair. But it means “more thymosin beta-4” is not an obviously safe research hypothesis in an inflamed joint.

Human trials of thymosin beta-4 do exist, just not in arthritis. Intravenous synthetic thymosin beta-4 at 42 to 1,260 mg daily for 14 days was well tolerated in healthy volunteers, and eye drops have been tested in severe dry eye. A 2026 review for orthopaedic surgeons in the American Journal of Sports Medicine concluded that human orthopaedic data for thymosin beta-4 and its derivative TB-500 are lacking, and that both remain banned in sport.

KPV, GHK-Cu and VIP: promising mechanisms, borrowed evidence

These three are frequently described as anti-inflammatory or regenerative options for joints. In each case the arthritis data either belong to a related molecule or stop at the animal or cell stage.

KPV is the three-amino-acid tail of alpha-melanocyte-stimulating hormone (α-MSH). The arthritis result belongs to the parent hormone: in rats with adjuvant arthritis, α-MSH given twice daily markedly inhibited the clinical and histological signs of disease, while prednisolone prevented the arthritis but caused marked, progressive weight loss. A review in Annals of the Rheumatic Diseases attributes most of α-MSH's anti-inflammatory activity to that C-terminal KPV sequence. KPV's own in vivo work, however, is in mouse models of colitis, where it reduced inflammation and aided recovery.

GHK-Cu, the copper tripeptide, is best known for skin and wound research. Its cartilage evidence is a 1995 cell-culture study in which it stimulated proliferation and collagen synthesis in chick embryo chondrocytes. The 2026 orthopaedic review found no clinical data supporting its use for musculoskeletal conditions.

VIP (vasoactive intestinal peptide) produced perhaps the most dramatic result of all. In a mouse model published in Nature Medicine, VIP significantly reduced the incidence and severity of experimental arthritis, completely abrogating joint swelling and destruction of cartilage and bone, by damping both the autoimmune and inflammatory components. It remains a striking animal finding.

A peptide that reached phase 3 — and didn't deliver

Calcitonin is the cautionary tale every peptide researcher should know. Oral salmon calcitonin was developed as a potential structure-modifying treatment for osteoarthritis, with markers of both bone and cartilage breakdown tracked alongside symptoms. It was taken all the way to two phase 3 trials enrolling 1,176 and 1,030 people with painful knee osteoarthritis, who took 0.8 mg of oral salmon calcitonin twice daily or placebo for 24 months.

Neither trial showed a significant effect on joint-space narrowing. One showed movement in WOMAC scores and in markers of bone and cartilage breakdown, but because of the trial's pre-specified hierarchical testing, that effect couldn't be claimed as statistically significant. The authors' conclusion was blunt: this formulation did not provide reproducible clinical benefits. Promising mechanisms, positive biomarkers and a sound rationale were not enough — which is exactly the gap between most research-stage peptides and the clinic.

How to read the arthritis peptide literature

Two reviews published in 2026 reached the same place from different directions. A scoping review of BPC-157, TB-500, CJC-1295, MK-677, ipamorelin and GHK-Cu found 67% of publications used preclinical animal models, that human studies were few and mostly lacked robust controls, and that the human data showed modest improvements at best for degenerative knee pain. The orthopaedic primer added that the indications, dosing, frequency and duration for these peptides remain unknown. When you read any new study, five questions do most of the work:

• What is the model? Adjuvant arthritis, collagen-induced arthritis and tendon transection are inflammatory, autoimmune and injury models. None of them is human osteoarthritis.
• Was there a control group? STEP 9's placebo arm improved by 27.5 WOMAC points. Without a comparator, that effect gets credited to the peptide.
• Was the outcome validated? WOMAC, VAS and joint-space width are standard. Recalled pain over the phone is not.
• How long did it run? Arthritis is chronic. Calcitonin needed 24 months to show it didn't slow structural change.
• Where does the evidence actually sit? For now: semaglutide and collagen have randomised human trials; BPC-157 has extensive animal data and one uncontrolled chart review; thymosin beta-4, KPV, GHK-Cu and VIP have animal, cell or biomarker data but no arthritis trials.

That doesn't make the research-stage compounds uninteresting — VIP's result in mice and BPC-157's consistency in rat injury models are exactly the kind of signals that justify further work. It means being precise about what has been shown, in which species, and in which disease.