Collagen peptides + vitamin C

Type I collagen is a triple helix of Gly-X-Y repeats where X is often proline and Y is often 4-hydroxyproline. Glycine is roughly one residue in three. Hydroxylation of those prolines is carried out by prolyl-4-hydroxylase, an enzyme that needs ascorbate to keep its iron centre reduced. Without vitamin C the helix never stabilises, which is scurvy described in a single sentence.

Now the part the label gets wrong. Digestion does not preserve provenance. Hydrolysed collagen is cleaved into free amino acids and small di- and tripeptides, and nothing routes them to your patellar tendon because they used to be a cow's. You do not build collagen out of collagen. You build it out of amino acids, like everything else.

What supplementation can plausibly do is deliver an unusual amino acid profile. Glycine and proline are non-essential, but for a matrix that is one-third glycine they may be conditionally rate-limiting during heavy synthesis. On top of that, hydroxyproline-containing dipeptides (Pro-Hyp, Gly-Pro-Hyp) survive digestion intact, appear in plasma, and are chemotactic for fibroblasts in vitro. That is a real, specific, testable mechanism, and it is far better than the one being advertised.

The other half of the mechanism is mechanical. Tendon collagen synthesis is gated by strain: tenocytes upregulate synthesis in response to loading, and a substrate pool only matters if the tissue has been told to build something. Hence the shape of the protocol used in the literature, which is a dose 30 to 60 minutes before short, heavy, intermittent loading, so that circulating amino acids peak while the strain signal is live.

Blunt caveat: after a loading bout both collagen synthesis and degradation rise. Net accretion is not a given, and the surrogate markers most studies use (serum PINP, for example) measure whole-body turnover rather than what happened in the one tendon you care about.

The founding study of the modern protocol is Shaw et al. 2017: eight healthy men, crossover design, 5 versus 15 g of gelatin with 50 mg vitamin C before jump-rope loading. It found a dose-dependent rise in a blood marker of collagen synthesis, and separately showed increased mechanical strength in engineered ligaments in a dish incubated with the subjects' serum. That is elegant work and a genuine mechanistic result. It is also n = 8 with no human tendon outcome.

Praet et al. 2019 layered specific collagen peptides onto standard eccentric rehab in chronic Achilles tendinopathy, n = 20, and found faster improvement on the VISA-A score. Encouraging, single-blinded, small. Clark et al. 2008 ran 147 athletes for 24 weeks on 10 g of collagen hydrolysate and reported reduced joint pain, with self-reported outcomes and funding from the ingredient manufacturer.

The skin and joint-comfort literature is larger and more consistent than the tendon literature, and it is also almost entirely industry-funded. It is not why athletes buy this, so it should not be what sells it to them.

What is missing is the whole ballgame: no adequately powered trial uses injury incidence, tendon structure on imaging, or return-to-play as its primary endpoint. And almost none include a control matched for total protein, so it remains genuinely unclear how much of the effect is collagen peptides rather than more protein and more vitamin C in an athlete who was short of both.