Healing & Recovery · 9 min read · Published September 8, 2026

Peptides for Joint Pain: BPC-157, TB-500, Lorecivivint, Pentosan and Collagen Peptides, Evidence Tiered

What the human trials say about joint pain candidates: lorecivivint (NDA under FDA review since January 2026 after two Phase 3 misses on pain), oral collagen peptides (three small RCTs), pentosan polysulfate (a 538-patient Phase 3 reading out in 2027), hyaluronic acid injections (89 trials, tiny effect), and the animal-only pair BPC-157 and TB-500.

Which joint pain peptides have human evidence?

Only two have human randomized trials: lorecivivint (Phase 3 trials of 498 and 513 patients, both missing their week-12 pain endpoint; NDA filed January 6, 2026) and oral collagen peptides (RCTs of 147 and 139 athletes). Pentosan polysulfate has a pilot RCT and a Phase 3 reading out in 2027. BPC-157 and TB-500 have animal data only.

Regulatory status: none of the five is an approved treatment for joint pain. Lorecivivint has a New Drug Application under FDA review. BPC-157 and TB-500 are unapproved; the July 23, 2026 advisory vote of 8 to 6 in their favor, against FDA staff advice, is non-binding (see the FDA status tracker). Collagen peptides are food supplements. Injectable pentosan polysulfate is investigational. Each is tiered below by what has been done in people, with hyaluronic acid injections as the comparator.

Is lorecivivint going to be approved for knee osteoarthritis?

Unknown. Biosplice submitted an NDA on January 6, 2026, citing 11 trials, over 1,800 patients dosed, and two-year joint-space-width preservation in OA-07. Both symptom trials missed their primary endpoint: OA-10 (498 patients, 28 weeks) and OA-11 (513 patients, 56 weeks; pain change minus 2.24 versus minus 2.49 on placebo, p equals 0.185). No review date is public.

Lorecivivint (SM04690) is not a peptide. It is a small-molecule CLK and DYRK1A kinase inhibitor that modulates Wnt signaling, given as a single 0.07 mg knee injection in the Phase 3 program. OA-10 randomized 498 patients with Kellgren-Lawrence grade 2 to 3 knee osteoarthritis and 'failed to meet the primary endpoint' of pain at week 12, with a post-hoc signal in the milder grade 2 subgroup (Clinical and Experimental Rheumatology, OA-10). OA-11 randomized 513 patients for 56 weeks; pain at week 12 fell 2.24 points on lorecivivint and 2.49 on placebo, and neither group lost meaningful joint space over 52 weeks, so the structural endpoint could not separate them either (PubMed 39808286). Evidence tier: human RCT, negative on symptoms.

The NDA rests instead on OA-07, a two-year Phase 3 in which patients given annual injections 'maintained and ultimately improved their medial JSW', and placebo patients who crossed over also gained joint space (Biosplice press release, January 6, 2026). If the FDA accepts a structural endpoint without a symptom win, lorecivivint would be the first disease-modifying osteoarthritis drug. That is a large 'if', and the point for peptide comparisons: it has done the trials the peptides have not, and may still not clear the bar.

Does BPC-157 work for joint pain in humans?

There is no controlled human study. A 2025 systematic review in HSS Journal found 35 preclinical studies and one human report: a retrospective series in which 7 of 12 patients given an intra-articular BPC-157 injection for chronic knee pain reported relief lasting more than six months. Rat tendon, ligament and muscle models are consistently positive. Evidence tier: animal.

BPC-157's musculoskeletal reputation comes from rat work by the Zagreb group and from a study of rat Achilles tendon fibroblasts showing faster explant outgrowth, better survival under oxidative stress and dose-dependent migration via the FAK-paxillin pathway (Journal of Applied Physiology, 2011). The HSS Journal review, which searched 1993 to June 2024, found no clinical safety data at all and only the single retrospective knee series (PubMed 40756949). It reports a half-life under 30 minutes, which leaves open how a peptide cleared that fast produces durable joint effects; nobody has measured it in people.

Pentadeca arginate, sold as a legal BPC-157 substitute, is the same sequence with a different salt and has no studies of its own. GHK-Cu turns up in 'joint stacks' such as KLOW and Wolverine blends; its evidence is topical, for skin, and we found no joint trial in any species.

What about TB-500 and thymosin beta-4?

TB-500 is a synthetic fragment of thymosin beta-4, a 43-amino-acid protein involved in actin regulation and wound repair. The joint-relevant evidence is animal: in a 2013 rat study, 1 microgram of thymosin beta-4 in fibrin sealant improved the structure and strength of a healing medial collateral ligament at four weeks. No human joint or tendon study of TB-500 exists.

The rat ligament study is the cleanest joint-adjacent data point for this family: treated ligaments showed evenly spaced collagen bundles, thicker fibrils and better biomechanical properties than controls at four weeks (Regulatory Peptides, 2013). Note what it used: full-length thymosin beta-4, placed directly in the wound in a sealant. Vials sold as TB-500 contain a shorter fragment injected elsewhere in the body. That gap between model and product runs through every compound in this article's bottom tier.

Do collagen peptides help joint pain?

Modestly, in three small RCTs. Penn State athletes taking 10 g of collagen hydrolysate daily for 24 weeks (147 randomized, 97 evaluable, 2008) improved on six pain measures versus placebo. In 139 athletes taking 5 g daily for 12 weeks (2017), activity pain fell 19.5 versus 13.9 mm. A 30-person knee osteoarthritis pilot showed MRI changes but no symptom change.

Oral collagen peptides are the odd member of this list: digested into small peptides and amino acids, regulated as food, and backed by more randomized human data than anything injectable below lorecivivint. The 2008 Penn State study randomized 147 varsity and club athletes with activity-related joint pain and no joint disease to 10 g collagen hydrolysate or placebo daily for 24 weeks; among the 97 with evaluable data, pain when walking, standing, at rest, carrying and lifting improved more on collagen, by roughly half a point on 10-point scales (Current Medical Research and Opinion, 2008). The 2017 study randomized 139 active adults with functional knee pain to 5 g bioactive collagen peptides or placebo for 12 weeks and found a 19.5 versus 13.9 mm drop in activity-related pain (p equals 0.046) (Applied Physiology, Nutrition, and Metabolism, 2017). Evidence tier: human RCT, small effect.

For osteoarthritis rather than athletic knee pain, the data are thinner. A 30-subject single-center pilot in mild knee osteoarthritis measured cartilage proteoglycan by delayed gadolinium-enhanced MRI over 24 weeks and found scores rising in two tibial regions on collagen while falling on placebo, with no difference in symptoms (Osteoarthritis and Cartilage, 2011). An imaging signal without a symptom signal in 30 people is a reason for a bigger trial, not a conclusion.

Where does pentosan polysulfate stand?

Between pilot and pivotal. A 2005 Perth RCT of 114 knee osteoarthritis patients (3 mg/kg intramuscular weekly for 4 weeks, followed 24 weeks) found differences on 3 of 7 clinical measures. A 2026 Phase 2 in 61 participants cut the cartilage-degradation marker ARGS 56.6 percent at day 56. The 538-patient Phase 3 reads out in the first quarter of 2027.

Pentosan polysulfate sodium is a semi-synthetic sulfated polysaccharide, not a peptide. The 2005 pilot showed shorter morning stiffness and lower pain at rest at several time points (Current Therapeutic Research, 2005); evidence tier: human RCT, pilot. The 2026 exploratory Phase 2 randomized 61 participants to twice-weekly, once-weekly or placebo subcutaneous injections for 39 days, with the ARGS reduction the one significant biomarker change (Arthritis Research and Therapy, 2026). The pivotal PARA_OA_012 trial gives 2 mg/kg subcutaneously twice weekly for six weeks across 65 sites, with a primary endpoint of average daily pain at day 112 (Paradigm Phase 3 enrollment report).

How do hyaluronic acid injections compare?

Poorly, once trial quality is accounted for. The 2012 Annals of Internal Medicine meta-analysis of 89 trials and 12,667 adults found a moderate overall pain effect (effect size minus 0.37) that shrank to a clinically irrelevant minus 0.11 in the 18 large trials with blinded outcome assessment, and a 1.41-fold higher risk of serious adverse events across 14 trials.

Viscosupplementation is the widely used injectable joint treatment that peptide marketing implicitly competes with. The Rutjes meta-analysis is a lesson in how small, unblinded trials inflate an effect: the five unpublished trials it located showed no effect at all (PubMed 22868835). Evidence tier: human RCT, many trials, minimal effect.

The bottom line

The two peptides with the loudest joint-repair marketing, BPC-157 and TB-500, have the least human evidence on this page. Lorecivivint ran the biggest trials, twice failed to beat placebo on pain, and is asking the FDA to approve it on cartilage measurements instead. The unglamorous oral collagen peptide has the most consistent, if modest, human signal. And the injectable comparator barely works when trials are blinded. Placebo responses to joint injections are large, which is why a 12-patient uncontrolled series of injected BPC-157 cannot tell you anything. For the safety side, see peptide side effects by class.