Transcript
Collagen is the most abundant protein in the human body. It holds skin together, cushions joints, and reinforces tendons. That biological centrality is why the supplement industry built a multibillion-dollar market around it. The reasoning sounds straightforward: your body makes less collagen as you age, so supplying it from outside should help. But the research behind that logic is far more complicated than the label suggests, and a 2025 meta-analysis in the American Journal of Medicine has exposed a structural problem in how that research has been produced.
The core issue is funding. When you look at all the skin trials combined, collagen appears to work. Skin hydration improves, elasticity rises, wrinkles diminish — the numbers are statistically significant across a pooled analysis of 23 randomised controlled trials involving 1,474 participants. That is a real body of evidence. But when researchers at South Korea's National Cancer Center and Monash University asked who paid for these studies, the picture changed completely.
By the authors' own classification, only five of those 23 trials were free of industry funding. When Myung and Park, the authors of that 2025 analysis, stratified results by funding source, the skin benefits in industry-funded trials held. In independently funded trials, the benefit on every single outcome — hydration, elasticity, wrinkles — disappeared. Not weakened. Disappeared. The same pattern held when studies were sorted by methodological quality: high-quality studies showed no benefit; only low-quality ones did, and only for elasticity.
This is not an unusual pattern in nutrition science. A 2007 analysis in PLOS Medicine of nutrition studies on soft drinks, juice and milk found that industry-funded work had roughly four to eight times the odds of reaching a conclusion favourable to the sponsor. Collagen appears to sit squarely inside that pattern.
The Collagen Stewardship Alliance, an industry group, has pushed back. Their argument is that Myung and Park misclassified some industry-backed trials as independent — by the Alliance's count, four of the five — which would make the independent subsample even smaller and statistically unreliable. It is a legitimate methodological challenge. But notice what that argument actually implies: if the independent evidence base is too small to draw firm conclusions in favour of collagen, it is also too small to have supported the confident positive claims that have been made for years. Pointing out that we have almost no clean independent data is not a defence of the supplement — it is an admission that the field has been running almost entirely on commercially interested research.
The honest summary of the skin evidence is this: twenty-three randomised controlled trials, 1,474 participants, and when you remove industry money from the equation, the effect on skin aging is no longer statistically significant. That is what the most funding-stratified collagen meta-analysis to date found.
Before treating that as proof the supplement does nothing at all, it is worth understanding the mechanism, because the mechanism is more interesting than either the industry or the critics tend to acknowledge.
Hydrolyzed collagen is collagen that has been broken down into smaller peptides — fragments of the original protein. The older critique of this category was simple and, if it held, devastating: you swallow protein, your stomach and gut break it into amino acids, it gets used wherever the body needs amino acids, and there is nothing special about it reaching your skin. Under that model, a collagen supplement is just an expensive and nutritionally incomplete protein powder.
That critique is partly right but oversimplified. Collagen is genuinely low in essential amino acids — it is not a complete protein in the conventional nutritional sense. But its concentrations of glycine, proline, and hydroxyproline are uniquely high, and those amino acids are not easily obtained at similar levels from standard dietary proteins. Some researchers have raised the possibility that observed benefits, where they do appear, might reflect correction of a low-protein intake rather than any specific collagen signalling mechanism. That remains an open question.
What is clearer now is that hydrolyzed collagen is not simply demolished in digestion. Specific bioactive peptides — including one called Gly-Pro-Hyp — are detectable in human blood after supplementation, and have been traced to skin tissue in animal studies. Research summarised in a 2021 systematic review by Khatri and colleagues found that these peptides survive the digestive process and reach systemic circulation. In cell culture experiments, they stimulate fibroblasts — the cells that produce collagen in the skin.
That is biologically plausible. The peptides survive. They reach tissue, at least in animals. They activate the relevant cells in a dish.
But the evidence stops short of the clinic. No independent human trial has yet connected those plasma peptide levels to a measurable, meaningful change in how skin looks or behaves. The mechanistic chain — peptide survives digestion, enters bloodstream, reaches skin, stimulates fibroblast, increases collagen production, changes skin appearance — is unbroken in cell culture and partially traced in humans, but the final clinical link in living people has not been independently confirmed. A 2024 study published in Frontiers in Nutrition did confirm plasma detectability of collagen peptides post-supplementation, but that study was funded by Rousselot BV, a major collagen manufacturer. Even the mechanistic supporting data carries commercial fingerprints.
To restate that clearly: the peptides do survive digestion and do reach the bloodstream — that is not in dispute. What is in dispute, and what the independent trial evidence cannot currently confirm, is whether that survival translates into any real benefit for human skin. The plausible mechanism and the clinical proof are not the same thing, and treating one as the other is where the marketing claims outrun the science.
Now, the joint story — and this is where the conversation genuinely shifts.
A 2024 trial-sequential meta-analysis published in Osteoarthritis and Cartilage looked at 35 randomised controlled trials involving 3,165 patients. It found small-to-moderate reductions in pain at moderate certainty and improvements in joint function at high certainty. This is a larger evidence base than the skin literature. The funding-source stratification that reshaped the skin evidence has not been systematically applied to the joint literature yet, so some caution is warranted. But the pattern of evidence is meaningfully different.
Within the joint research, one trial included in the Khatri review, by Zdzieblik and colleagues in 2017, gave athletes with activity-related knee pain five grams a day of collagen peptides for twelve weeks, and found less pain during activity than with placebo — a real difference, though a modest one.
Before that joint evidence is taken as settling anything, though, there is a product distinction that matters enormously and is almost never made clearly in supplement marketing.
The joint market includes something called UC-II — undenatured type II collagen — and it is a completely different product operating by a completely different mechanism. Treating it as interchangeable with hydrolyzed collagen is one of the more persistent misunderstandings in this space.
Hydrolyzed collagen peptides are thought to act as raw material. The logic is supply-side: give the body the building blocks it needs to produce or repair collagen in joints and cartilage. UC-II does not work that way at all. It is intact, undenatured collagen — collagen that has not been broken down — and it is used in doses around 40 milligrams per day, which is far too small to provide any meaningful supply of raw materials. That small dose is intentional.
The proposed mechanism for UC-II is oral immune tolerance. The idea is that exposing the gut's immune system to intact cartilage collagen trains it to tone down inflammation directed at cartilage collagen. The idea borrows from autoimmune joint disease such as rheumatoid arthritis; osteoarthritis is mainly degenerative, so there the proposed benefit is damping the inflammation that accompanies cartilage breakdown. In a trial documented by Khatri and colleagues, UC-II at approximately 40 milligrams per day was associated with longer exercise before knee discomfort set in and improved knee extension. The oral tolerance mechanism has not been definitively confirmed in large-scale human trials, but the findings are distinct enough to merit separate consideration from the hydrolyzed peptide literature.
The distinction is worth stating plainly. Hydrolyzed collagen: broken down, larger doses, supply-side logic, building-block mechanism. UC-II: intact, small doses, immune-modulation logic, oral tolerance mechanism. These are not two versions of the same product. They are two different interventions that happen to share the word collagen. Conflating them — or drawing conclusions from one as though they apply to the other — produces exactly the kind of confusion that supplement marketing exploits.
To restate the two-product distinction before the closing: collagen peptides marketed for skin or joint support are operating on the supply-side, raw-material logic — and the skin evidence for that logic is almost entirely industry-funded, while the joint evidence is stronger but not yet subjected to the same funding-source scrutiny. UC-II is claiming to work through immune tolerance at a dose too small to function as a building-block supplement — a mechanistically distinct claim that requires distinct evidence to evaluate.
The broader lesson is about the structure of the evidence, not just the conclusions. The collagen skin literature has a fundamental problem: it has been built almost exclusively by people with a financial stake in the outcome. The 2025 Myung and Park analysis is notable not because it proves collagen does nothing, but because it is the first analysis to cleanly ask the funding question at scale and to find that the answer changes everything. Eighteen of twenty-three trials industry-funded. No significant benefit in the independent subset. That is the evidentiary reality behind a multibillion-dollar market.
The joint literature is larger and shows higher certainty — that is the clearest and most grounded distinction the current evidence supports. The 2024 Osteoarthritis and Cartilage meta-analysis, with 35 trials and over three thousand patients, found moderate certainty for pain reduction and high certainty for functional improvement. Those numbers have not been subjected to the same funding-source audit that changed the skin numbers, so the joint case is stronger on paper but has not yet been stress-tested in the same way.
What remains genuinely unresolved? Several things. The number of truly independent, adequately powered skin trials is small enough that even the Myung and Park finding — the independent subsample showing no significant benefit — may not be the final word. If there were ten well-run, well-funded, independent skin trials instead of five, we would know considerably more. The cell-culture evidence that bioactive peptides stimulate fibroblasts has not been translated into confirmed clinical benefit in humans at typical supplement doses. The funding-source lens that reshaped the skin literature has not been systematically applied to joint research. The oral tolerance mechanism behind UC-II is biologically coherent and supported by some trial data, but has not been confirmed at scale in humans. That does not make it wrong. It makes it unconfirmed.
Collagen is two products wearing one label, aimed at different targets by different mechanisms, supported by evidence of very different quality. For skin, the positive signal in the literature tracks commercial interest more reliably than it tracks the supplement's effects. For joints, the evidence is more substantial, though it has not yet been audited for funding and has its own gaps. And the absorption story — peptides surviving digestion and reaching tissue — is real, but the distance between a detectable plasma peptide and a clinically confirmed improvement in skin, measured in a well-run independent trial, has not yet been closed.