Injuries & Orthopedics
Best Peptides for Osteoarthritis: Evidence & Safety (2026)
An evidence-graded review of the peptides marketed for osteoarthritis and joint degeneration. The honest 2026 verdict: the strongest human data belongs not to any joint-repair peptide but to GLP-1 metabolic drugs that relieve knee-OA pain by removing mechanical load.
GLP-1 agonistsPentosan polysulfateBPC-157AOD-9604GHK-Cu
The quick verdict
Ranked by strength of evidence for osteoarthritis specifically — and the honest verdict is that the top-tier human data belongs to GLP-1 metabolic drugs that relieve knee-OA pain by removing mechanical load, not to any peptide shown to regrow cartilage.
- Best overall
- GLP-1 receptor agonists (semaglutide, retatrutide) — The only OA candidate with Grade A human data — two Phase 3 RCTs (STEP 9, TRIUMPH-4) met co-primary WOMAC pain endpoints — but the proven mechanism is weight loss and load reduction, not cartilage repair.
- Best value
- Weight / load management + rehabilitation (non-peptide baseline) — Obesity is the single most modifiable driver of knee OA. The interventions with the strongest human evidence — weight loss, load management and structured rehab — are exactly what the GLP-1 data validates, and they are what any peptide should be measured against.
- Best for Interest in a joint-targeted disease-modifying drug (DMOAD)
- Pentosan polysulfate (PPS) — PPS (Grade B; not a peptide) is the leading DMOAD candidate — a positive pilot RCT for pain and stiffness plus an open trial showing a fall in a cartilage-degradation biomarker, with Phase 3 subcutaneous trials active — though structural cartilage preservation remains unproven.
How we evaluated
We ranked each candidate strictly by the strength of published evidence for osteoarthritis specifically — not by marketing volume or general popularity. We separated human RCT data from lower-tier human, animal and in-vitro data; weighted on-target OA models and endpoints (WOMAC pain, cartilage biomarkers, knee-OA models) above generic tissue-repair claims; distinguished pain/function benefit from proven structural disease modification; and graded honestly where the only support is preclinical, mechanistic or promotional. Only the GLP-1 class reaches Grade A, and it does so for pain via weight loss, not cartilage regrowth.
- On-target human evidence. Completed randomized, controlled human trials in osteoarthritis. Only the GLP-1 class (STEP 9, TRIUMPH-4) qualifies at Grade A; pentosan polysulfate has a pilot RCT plus an open trial (Grade B); the joint peptides have none.
- Pain/function vs structural disease modification. Whether the evidence shows symptom relief (pain, WOMAC, function) or actual cartilage preservation/regrowth on imaging or biomarkers. No agent here has completed structural-endpoint proof of cartilage regrowth in humans.
- On-target preclinical evidence. Animal and in-vitro models specific to OA cartilage — knee-OA models, chondrocyte GAG synthesis, MMP downregulation — weighted above generic wound or gut models.
- Mechanistic plausibility and delivery. Whether the proposed mechanism targets the rate-limiting biology of OA — mechanical load, catabolic inflammation, matrix anabolism — and whether a signal can actually reach the avascular cartilage matrix at therapeutic concentration.
- Safety and legal status. Condition-specific risks (notably intra-articular infection and PPS bleeding/maculopathy), FDA approval and compounding status, and sport/military prohibition as of June 2026.
Rating scale: 1-5 stars reflecting strength of evidence FOR osteoarthritis specifically (5 = controlled human structural proof; 4 = controlled human pain/function proof; 3 = on-target preclinical only; 1 = no independent data). Only the GLP-1 class exceeds 3.
Last verified .
At a glance
| # | Name | Evidence | Rating | Best for | Pricing |
|---|---|---|---|---|---|
| 1 | GLP-1 Receptor Agonists (Semaglutide & Retatrutide) | A | 4.0 | Overweight patients with knee OA researching the best-evidenced peptide-class intervention — understanding the benefit is pain relief via weight loss, not cartilage repair | Semaglutide FDA-approved for weight/diabetes (not OA); retatrutide investigational, trials only |
| 2 | Pentosan Polysulfate (PPS) | B | 3.0 | Readers specifically interested in the leading disease-modifying-OA-drug candidate and the state of DMOAD trials — while noting PPS is not a peptide | Oral Elmiron FDA-approved for interstitial cystitis only; no approved injectable/OA form in the US |
| 3 | BPC-157 | C | 2.5 | Readers researching the most on-target preclinical peptide evidence for OA — understanding it remains unproven in humans and carries real intra-articular safety hazards | Not FDA-approved; sold as research chemical / via compounding gray zone |
| 4 | AOD-9604 (LAT8881) | C | 2.0 | Readers tracking the most cartilage-specific peptide mechanism and the LAT8881 OA program — understanding human joint efficacy is unproven | Not FDA-approved for any indication; sold as research chemical / via compounding gray zone |
| 5 | GHK-Cu (Copper Tripeptide) | C | 1.5 | Readers interested in cartilage-relevant peptide mechanisms in cell culture — understanding GHK-Cu has no human OA evidence and an unsolved delivery problem | Topical widely available; injectable not FDA-approved and in the compounding gray zone |
GLP-1 Receptor Agonists (Semaglutide & Retatrutide)
The only OA candidate with Grade A human data — but for pain via weight loss, not cartilage repair
GLP-1 receptor agonists are large synthetic peptide drugs — semaglutide is a once-weekly GLP-1 agonist, and retatrutide (LY3437943) is a first-in-class triple agonist of GIP, GLP-1 and glucagon. This is the single place where 'peptides for osteoarthritis' rests on Grade A human evidence, precisely because the target is metabolic rather than chondral. In STEP 9, a 68-week, double-blind, placebo-controlled Phase 3 RCT (n=407 adults with obesity and moderate knee OA), semaglutide 2.4 mg weekly produced −13.7% body weight versus −3.2% and a WOMAC pain reduction of −41.7 versus −27.5 points (treatment difference roughly −14 points, P<0.001), with greater physical-function improvement. In TRIUMPH-4 (Phase 3, n=445, topline December 2025), retatrutide 9 mg and 12 mg met all co-primary and key secondary endpoints, with weight loss up to −28.7% and WOMAC pain reduction up to −75.8% versus −40.3% for placebo. The decisive limitation: both trials measured pain and function, not joint structure. Neither demonstrates cartilage preservation or regrowth; the established mechanism is weight loss — load reduction plus reduced adipose-driven inflammation. Any direct chondroprotective effect independent of weight loss is hypothesis-generating, not proven. Semaglutide is FDA-approved for weight/diabetes, not OA; retatrutide is investigational and not approved.
Strengths
- The only OA candidate with Grade A human evidence — two Phase 3 RCTs (STEP 9, TRIUMPH-4) with co-primary WOMAC pain endpoints met versus placebo
- Targets the single most modifiable driver of knee OA — mechanical load from excess body weight — a genuine root-cause mechanism
- Large, sustained weight loss (up to ~14% semaglutide, up to ~29% retatrutide) that unloads the joint and reduces systemic inflammation
- Semaglutide is an FDA-approved drug with an established human safety profile (for its weight/diabetes indication)
Weaknesses
- Proven benefit is pain and function only — neither trial measured cartilage structure, so this is NOT proven disease modification or cartilage regrowth
- Retatrutide is investigational, not FDA-approved; compounded GLP-1s are unapproved, misbranded drugs per FDA warning letters
- Class GI effects (nausea, vomiting, diarrhea), pancreatitis caution, contraindicated with personal/family history of medullary thyroid carcinoma or MEN2; WADA-prohibited when used for performance
- Best for
- Overweight patients with knee OA researching the best-evidenced peptide-class intervention — understanding the benefit is pain relief via weight loss, not cartilage repair
- Pricing
- Semaglutide FDA-approved for weight/diabetes (not OA); retatrutide investigational, trials only
Pentosan Polysulfate (PPS)
Leading disease-modifying-OA-drug candidate — but it is not actually a peptide
Despite constant grouping with BPC-157 in joint-peptide lists, pentosan polysulfate is not a peptide — it is a semi-synthetic sulfated xylan polysaccharide, a heparinoid and glycosaminoglycan mimetic with no amino acids or peptide bonds. It is included because it is the leading true disease-modifying-OA-drug candidate marketed alongside peptides. Its human OA evidence reaches Grade B for pain and a biomarker, not proven structure. A randomized, double-blind, placebo-controlled pilot RCT (Ghosh 2005; n=114; sodium PPS 3 mg/kg intramuscularly weekly for four weeks versus Ringer's) found significant benefit over placebo for joint stiffness, rest pain and patient global assessment, persisting roughly 20 weeks after a four-week course. An open-label trial (Kumagai 2010; n=20; 2 mg/kg subcutaneously weekly for six weeks) reported a roughly 19–22% fall in serum C2C, a type-II-collagen cartilage-degradation biomarker, at weeks 11 and 24, with large pain reductions. Neither used MRI or radiographic structural endpoints, so cartilage preservation remains unproven and any structural claim is Grade C. Mechanistically PPS stimulates chondrocyte proteoglycan synthesis, inhibits cartilage-degrading enzymes, and suppresses IL-1β-driven inflammatory signaling; it is a long-established veterinary DMOAD (Cartrophen) in dogs and horses. Confirmatory Phase 3 subcutaneous trials are active in 2025–2026. The FDA-approved oral product, Elmiron, is for interstitial cystitis only.
Strengths
- The leading true DMOAD candidate here — a positive pilot RCT for pain/stiffness plus an open trial showing a fall in a cartilage-degradation biomarker (serum C2C)
- Coherent chondroprotective mechanism — stimulates proteoglycan synthesis, inhibits cartilage-degrading enzymes, suppresses IL-1β/NF-κB signaling
- Active Phase 3 subcutaneous human trials specifically in knee OA (NCT06917404, NCT04809376)
- Long-established veterinary DMOAD (Cartrophen) with a track record in dogs and horses
Weaknesses
- Not actually a peptide — a common category error that inflates its association with joint peptides
- No completed trial with MRI/radiographic structural endpoints, so human cartilage preservation is still unproven (structural claim only Grade C)
- A weak heparinoid that prolongs bleeding, plus a serious cumulative-dose-dependent, potentially irreversible pigmentary maculopathy (FDA warning); no FDA-approved human injectable in the US and no approved OA indication
- Best for
- Readers specifically interested in the leading disease-modifying-OA-drug candidate and the state of DMOAD trials — while noting PPS is not a peptide
- Pricing
- Oral Elmiron FDA-approved for interstitial cystitis only; no approved injectable/OA form in the US
Source: Ghosh et al., pilot RCT, Curr Ther Res 2005 (PMID 24678076)
BPC-157
Broadest preclinical OA data in the class — but one uncontrolled human series, no controlled trial
BPC-157 is a synthetic stable gastric pentadecapeptide (sequence GEPPPGKPADDAGLV) derived from a human gastric protein, and it carries the broadest preclinical OA case of the true joint peptides — though heavily single-lab-dominated. In a rat knee-OA model (ACL/MCL transection plus medial meniscectomy), controls developed near-total cartilage and proximal-tibia destruction by eight weeks, whereas intra-articular BPC-157 rats had articular surfaces similar to non-operated animals at four weeks, only minor lesions at eight weeks, preserved bone on X-ray, and restored gait, leg-pressure and knee-mobility function. Mechanistically it upregulates VEGFR2 and modulates the nitric-oxide (Akt–eNOS) pathway, promoting capillary ingrowth and organized collagen in healing soft tissue — plausibly relevant at vascular peri-articular tissue but of little leverage inside avascular hyaline cartilage, the tissue OA destroys. The weakness is decisive: a 2025 systematic review of BPC-157 in orthopaedic sports medicine included 36 studies, 35 preclinical and only 1 clinical. The only OA-relevant human report is a retrospective, uncontrolled case series of intra-articular BPC-157 for assorted knee pain, with no control group, no validated outcomes, no imaging, and a single author and journal. Because all published human BPC-157 reports share the same lead author, they cannot serve as independent confirmation. No controlled human OA trial exists. Honest grade: C — best preclinical breadth in the class, but far from proven.
Strengths
- The broadest on-target preclinical OA data of the true joint peptides — a rat knee-OA model that preserved cartilage, bone and gait versus near-total control failure
- Coherent tissue-repair mechanism (VEGFR2 upregulation, nitric-oxide modulation, organized collagen) relevant to vascular peri-articular soft tissue
- The only OA-relevant human data among the classic joint peptides exists at all (an uncontrolled intra-articular case series)
Weaknesses
- Zero completed randomized, placebo-controlled human OA trials — the only human data is an uncontrolled, single-author case series with no imaging or validated outcomes
- Its angiogenic mechanism has little leverage inside avascular hyaline cartilage, the exact tissue OA destroys
- Unapproved drug in an FDA compounding gray zone; prohibited in sport (WADA S0) and on the DoD banned list; unregulated product risks septic arthritis on intra-articular injection
- Best for
- Readers researching the most on-target preclinical peptide evidence for OA — understanding it remains unproven in humans and carries real intra-articular safety hazards
- Pricing
- Not FDA-approved; sold as research chemical / via compounding gray zone
Source: Vasireddi et al., systematic review, HSS Journal 2025 (PMID 40756949)
AOD-9604 (LAT8881)
The most cartilage-specific mechanism — but the only OA data is a single rabbit study
AOD-9604 is a 16-amino-acid synthetic fragment of the C-terminal lipolytic tail of human growth hormone (Tyr-hGH 177–191), originally an anti-obesity candidate and re-coded LAT8881 for a cartilage/OA program. An independent 2025 orthopaedics review singles it out as receiving the most specific cartilage evidence of the candidate peptides — but that evidence is a single animal study. In a collagenase-induced knee-OA model in 32 rabbits, weekly intra-articular AOD-9604 (0.25 mg), especially combined with hyaluronic acid (6 mg), enhanced cartilage regeneration and reduced histopathologic degeneration versus saline. It is proposed to act via lanthionine synthetase C-like protein (LANCL) activation, a mechanism explored specifically for cartilage — but this remains preclinical. The human evidence is the weakness: AOD-9604 completed six human RCTs (n=893), but all for obesity, all establishing only safety, with no orthopaedic RCT. Any human OA claim is therefore an extrapolation from one rabbit study. Obesity-RCT systemic safety was described as indistinguishable from placebo, with no IGF-1 or glucose signal, but intra-articular human safety is entirely uncharacterized, and gray-market product carries identity and sterility risk for joint injection. Honest grade: C for OA — the most cartilage-specific mechanism on paper, with essentially no human joint data.
Strengths
- Singled out by an independent 2025 orthopaedics review as having the most cartilage-specific mechanism of the candidate peptides
- The one animal OA study (rabbit, intra-articular ± hyaluronic acid) showed enhanced cartilage regeneration versus saline
- Human obesity RCTs (n=893) found systemic safety indistinguishable from placebo, with no IGF-1 or glucose signal
Weaknesses
- The entire human OA case is extrapolation from a single rabbit study — six human RCTs exist but all for obesity, establishing safety only, with no orthopaedic RCT
- Intra-articular human safety is entirely uncharacterized; the LANCL cartilage mechanism is preclinical
- Unapproved for any indication; prohibited in sport (WADA S2.2, growth-hormone fragment); gray-market product carries identity/sterility risk for joint injection
- Best for
- Readers tracking the most cartilage-specific peptide mechanism and the LAT8881 OA program — understanding human joint efficacy is unproven
- Pricing
- Not FDA-approved for any indication; sold as research chemical / via compounding gray zone
Source: Kwon & Park, rabbit knee-OA model, Ann Clin Lab Sci 2015 (PMID 26275694)
GHK-Cu (Copper Tripeptide)
Strong human skin data, but the OA case is in-vitro chondrocyte signaling only
GHK-Cu is glycyl-L-histidyl-L-lysine bound to copper (1:1), a human matrix-remodeling tripeptide isolated by Pickart in 1973. It has genuinely strong human evidence — but for skin and wound healing, not joints. Its OA case is entirely in-vitro and mechanistic: at nanomolar (1–10 nM) concentrations it stimulates collagen, dermatan and chondroitin sulfate, and decorin synthesis; in stressed chondrocyte culture it raised glycosaminoglycan synthesis and reduced IL-1β-induced cell death; and it downregulates MMP-1 (roughly 37%) and MMP-3 (roughly 52%), enzymes relevant to OA cartilage catabolism. An independent orthopaedics review classifies its joint use as preclinical and early-studies only. No human knee or cartilage clinical trial supports OA efficacy. There is also a hard delivery problem: a peptide that boosts collagen and GAG in a dish must still reach cartilage at therapeutic concentration through the synovium, and collagen-turnover kinetics mean any effect would be slow, over weeks to months. Its use is well characterized topically, but injectable and intra-articular joint protocols are anecdotal clinic practice without controlled support, and injectable GHK-Cu was swept into the FDA compounding review. Honest grade: C/D for OA — a mechanistically interesting chondrocyte-signaling molecule with no human joint evidence.
Strengths
- Genuinely strong human evidence base — but for skin and wound healing, not joints
- Mechanistically relevant to OA cartilage in vitro — stimulates collagen/GAG synthesis and downregulates MMP-1 and MMP-3 in chondrocyte models
- Well-characterized safety profile when used topically
Weaknesses
- No human knee or cartilage clinical trial supports OA efficacy — the entire joint case is in-vitro and mechanistic
- Unsolved delivery: an in-vitro chondrocyte signal must still reach cartilage through the synovium at therapeutic concentration, with slow collagen-turnover kinetics
- Injectable/intra-articular joint safety is not established in controlled human studies; injectable GHK-Cu was swept into the FDA compounding review
- Best for
- Readers interested in cartilage-relevant peptide mechanisms in cell culture — understanding GHK-Cu has no human OA evidence and an unsolved delivery problem
- Pricing
- Topical widely available; injectable not FDA-approved and in the compounding gray zone
Frequently asked
Which peptide has the best evidence for osteoarthritis?
If you count GLP-1 receptor agonists as peptide drugs — and they are — the GLP-1 class has the strongest human OA data, Grade A, for knee-OA pain reduction via weight loss. Semaglutide in STEP 9 and retatrutide in Phase 3 TRIUMPH-4 both significantly cut WOMAC pain versus placebo. The crucial caveat: both trials measured pain and function, not joint structure, and the established mechanism is weight loss, not cartilage repair. For a joint-targeted disease-modifying drug, pentosan polysulfate leads (Grade B, and it is not a peptide), with Phase 3 trials ongoing. The classic joint peptides — BPC-157, AOD-9604 and GHK-Cu — have no controlled human OA efficacy data at all.
Do GLP-1 drugs like semaglutide actually heal the joint, or just help me lose weight?
The proven benefit is pain and function via weight loss. STEP 9 and TRIUMPH-4 did not measure cartilage structure, and a direct disease-modifying effect on cartilage independent of weight loss is unproven — hypothesis-generating only. From a root-cause standpoint that is still meaningful: obesity is the leading modifiable driver of knee OA, and each pound of body weight multiplies across the knee during gait, so unloading the joint addresses a genuine cause rather than only masking a symptom. But it is not evidence of cartilage regrowth. Whether GLP-1 signaling is chondroprotective in its own right would require structural-endpoint trials, which is what the tirzepatide STOP KNEE-OA study and future work are designed to test.
Can BPC-157 or AOD-9604 fix my arthritic knee instead of surgery?
No controlled human evidence supports that. BPC-157's joint data is a single-lab rat knee-osteoarthritis model plus one retrospective, uncontrolled human knee-pain case series with no imaging or validated outcomes. AOD-9604's OA data is a single rabbit study using intra-articular dosing with hyaluronic acid, on top of six human trials that were all for obesity and established only safety. Neither has a completed randomized human OA trial. Human articular cartilage is thicker, slower and far less regenerative than animal-model cartilage, so promising rodent and rabbit findings routinely fail to translate. Advanced OA decisions, including whether to consider joint replacement, belong with an orthopaedic surgeon.
Is pentosan polysulfate a peptide, and why is it in this review?
No — pentosan polysulfate is not a peptide. It is a semi-synthetic sulfated xylan polysaccharide, a heparinoid and glycosaminoglycan mimetic with no amino acids or peptide bonds. It is included because it is constantly grouped with BPC-157 in joint-peptide marketing and, more importantly, because it is the leading true disease-modifying-OA-drug candidate in this space. It has a positive pilot RCT for OA pain and stiffness, an open-label trial showing a fall in a type-II-collagen cartilage-degradation biomarker, and active Phase 3 subcutaneous trials. The oral form, Elmiron, is FDA-approved only for interstitial cystitis, not joints, and carries a serious cumulative-dose retinal-toxicity warning.
Can I buy retatrutide for my knees now, and will these fail a drug test?
No, you cannot legally buy retatrutide for OA. Retatrutide is investigational, not FDA-approved, and available only through clinical trials; compounded versions are unapproved, misbranded drugs per FDA warning letters. For athletes the sport picture is strict: retatrutide is WADA-prohibited at all times under S0, AOD-9604 is prohibited under S2.2 as a growth-hormone fragment, and BPC-157 and TB-500 are prohibited year-round under S0, with BPC-157 also on the U.S. Department of Defense prohibited-ingredient list. Even approved GLP-1 use may require medical documentation. Anyone subject to testing should verify current status through GlobalDRO or USADA before using any substance discussed here.