Collagen and bone

Something worth being upfront about: the best structural evidence collagen supplementation has produced in humans is not about cartilage at all. It is about bone, and it is better than anything in the cartilage literature.

This entry exists so that result is recorded accurately rather than either ignored or quietly borrowed.

The trial

One hundred and thirty-one postmenopausal women with age-related low bone density — spine T-score −2.4, femoral neck −1.4 — randomized to 5 g of specific collagen peptides or placebo, daily, for twelve months, with DXA as the primary endpoint.

Spine T-score rose by 0.1 on collagen and fell by 0.03 on control (p = 0.030). Femoral neck rose 0.09 against −0.01 (p = 0.003). Both were prespecified as primary endpoints and both survive the trial's own multiplicity correction.

The trial's blood markers are usually quoted alongside those numbers and they carry less than they appear to. The bone-formation marker P1NP rose in the collagen group and the resorption marker CTX-1 rose in the control group, which sounds like two arms diverging. Each was tested only against its own group's starting value, though; the paper reports no comparison between the arms on either marker. The control group's P1NP was in fact the higher of the two both at the start and at twelve months.

Compare that with cartilage, where the entire imaging evidence base is one 30-person pilot in which four of six regions and every clinical outcome were null. A year-long trial in 131 people with a validated imaging endpoint is a different class of evidence, whatever the blood markers do or do not add.

The caveats

One trial, no independent replication. Everything downstream is the same group.

The conflict declaration deserves scrutiny. The paper states that all five authors declare no conflict of interest. Its acknowledgements, a few lines earlier, state that part of the study was financed by GELITA AG — the company whose branded peptide is the intervention — and one of the authors is at the GELITA-linked Collagen Research Institute in Kiel. This guide flags the same pattern in the UC-II literature, where a trial funded by the ingredient's owner also declared none. It does not make the data wrong. It does mean the declaration is not doing the work declarations are supposed to do.

The randomisation did not come out level. The collagen group started with significantly lower spine density than the control group, which is why the analysis is a covariance model rather than a plain comparison of change. Of the 131 women, 102 finished, and the missing twelve-month measurements were filled in by imputation. Both groups were encouraged to take calcium and vitamin D, and nobody prescribed or recorded how much.

The four-year follow-up is not a trial. Thirty-one of the original 131 women continued, all of them on the supplement, open-label, with no control arm. It reports continued gains in spine density. What it cannot do is separate those gains from selection — the 31 who stayed four years are unlikely to be a random sample of the 131.

The bone-stock footnote

There is a second, stranger piece of bone evidence, and it belongs to food rather than supplements.

Researchers fractionated chicken-vegetable bone broth by ion-exchange chromatography and tested each fraction on osteoclast differentiation. The fraction containing hyaluronan and chondroitin sulfate was the one that worked, suppressing osteoclast differentiation and its marker genes. In ovariectomised rats, both the whole broth and that fraction raised bone mineral density and bone volume relative to tissue volume.

Rats, and a bone endpoint. But it is one of only two animal studies of bone stock in existence, and unlike the other one it identified which constituents were responsible instead of assuming.

Bone against cartilage

Bone and cartilage are both collagen-based and that is roughly where the similarity ends.

Bone is type I collagen mineralised with hydroxyapatite, richly vascularised, and remodelled continuously by a coupled osteoblast–osteoclast system that responds to systemic hormonal signals. Articular cartilage is type II collagen in a proteoglycan gel, avascular, aneural, with famously slow matrix turnover and no equivalent remodelling unit. A supplement can plausibly shift the balance of an active remodelling cycle without doing anything for a tissue that barely turns over and has no blood supply to deliver anything through.

The bone result is also consistent with a mechanism that has nothing to do with cartilage: collagen peptides are a protein source, and protein intake independently affects bone density in older adults.

So this is a real finding, and it is not evidence for the joint claim. If anything it sharpens the contrast — when collagen has been given a long trial with a hard structural endpoint in a tissue that can respond, it produced a measurable result. Nobody has run that trial for cartilage.