Injury recovery: what the research says

Evidence limits

What has injury recovery research on BPC-157 and TB-500 not shown?

Replication is a second gap. A 2026 review in Pharmaceutics estimated that more than 80% of BPC-157 papers come from one research group at the University of Zagreb, and it called the lack of independent replication a significant limitation (Mateescu et al., 2026). For TB-500, the FDA adds that acetylation, the chemical tag on the fragment, changes how a peptide behaves, so findings on the full protein or the untagged piece cannot be directly extrapolated to TB-500.

Healing is hard to judge without a comparison. Tissue mends with time, and changes in care can also play a part, so a report with no control group cannot show what caused an improvement (Poolton et al., Acta Orthopaedica Belgica, 2026). That is why the rat studies below set treated animals beside untreated ones, and why their results still stop at the animal.

What did BPC-157 studies find after tendon and muscle injuries in rats?

BPC-157 tendon and muscle findings come from rat injury models. After the Achilles tendon was cut, treated rats had a higher load before failure and higher walking-function scores than saline-treated rats (Staresinic et al., Journal of Orthopaedic Research, 2003). After a complete cut of the quadriceps, a 2006 study by Staresinic and colleagues reported a higher breaking load and larger myofibril diameters in treated rats. Neither abstract gives a group size, and neither result shows the same in people.

Myofibrils are the contracting threads inside muscle cells. A crush-injury study in rats reported less haematoma, a pooled bruise of blood, and less oedema, or swelling, with better functional and enzyme-activity outcomes over a 14-day period (Novinscak et al., Surgery Today, 2008). Its abstract gives no group size. In a 2026 study of 32 rats with repaired Achilles tendons, four per group analysed mechanically, BPC-157 had higher maximum-load-to-failure values than control without statistical significance (Biçer et al., Joint Diseases and Related Surgery, 2026).

The FDA briefing and PubMed records searched on 14 October 2026 did not find a human study of BPC-157 at a tendon. The BPC-157 study findings guide gives each model, its result and its limit, and BPC-157 compared with TB-500 explains how the two molecules differ in structure and origin.

What did TB-500 studies find on wound and tendon healing?

TB-500 wound evidence is indirect: the FDA's May 2026 briefing reports no human study of the fragment. In a 2003 study of aged mice with skin wounds, its parent protein, thymosin beta-4, and an untagged fragment each increased skin closure on day 7, a mouse result that does not describe TB-500 (Philp et al., 2003). In a 2024 cell study, TB-500 itself did not close scratch wounds significantly more than vehicle, the liquid without it (Rahaman et al., 2024).

The 2003 mouse study appeared in Wound Repair and Regeneration, and the FDA's summary of it does not state the number of mice. The 2024 scratch test, in the Journal of Chromatography B, watched fibroblasts for 8 hours, and the FDA summary does not state the number of cultures. Both are early-stage results, one in animals and one in a dish.

The same 2026 rat tendon study that tested BPC-157 included material labelled TB-500, which reached statistical significance on maximum load to failure (Biçer et al., 2026). That is an exploratory animal result. The study does not independently confirm the material's molecular identity, and it does not establish a human result. The TB-500 findings guide separates each finding by the molecule that was tested.

Have BPC-157 or TB-500 been studied in people with injuries?

No controlled human injury study of BPC-157 or TB-500 appears in the sources cited here. For TB-500, the FDA found no information in which it was given to patients to treat any condition. For BPC-157, a retrospective knee chart review of 17 patients recorded relief of knee pain in 14 of the 16 reached by phone, with no scale, imaging or control group (Lee and Padgett, Alternative Therapies in Health and Medicine, 2021).

Poolton and colleagues rated the human evidence for BPC-157 as very weak in an October 2026 review in Acta Orthopaedica Belgica. The knee report records pain recalled by patients, rather than proving tissue repair. Registry record NCT07803250, from the University of Arkansas, lists a randomised Phase 1 pilot, an early-stage trial, of 30 planned participants after rotator cuff repair surgery. It was not yet recruiting on 14 October 2026, and a registry record is a plan, not a result.

The human wound trials in these sources tested the parent protein, full-length thymosin beta-4, and none used TB-500. In a pressure-ulcer trial of 72 people, registry record NCT00382174 lists 8 of 54 treated and 3 of 18 on placebo as healed at day 84, with no significance test. In a venous leg ulcer trial, record NCT00832091 lists wound closure without drainage at day 84 in 12 of 55 treated and 4 of 17 on placebo. In both trials, similar proportions of each group healed.