Collagen Peptides

01The triple helix and the Gly-X-Y repeat

Collagen is a right-handed supercoil of three parallel left-handed polyproline-II helices. That sentence is established. Shoulders and Raines restated it as the architectural fact from which every later chemical peculiarity follows: every third residue is glycine, the X and Y positions are often proline and 4-hydroxyproline, and the glycine at each turn is what lets the three chains pack (Shoulders and Raines, 2009). The same review is the source for the claim that types V and XI nucleate type I and type II fibrils. Kadler, Hill and Canty-Laird placed those nucleating collagens in the living fibril, together with fibronectin and integrins, and refused the cartoon in which type I self-assembles from a purified solution (Kadler, Hill and Canty-Laird, 2008). Kadler and Baldock’s glance map exists as a verified title (Kadler and Baldock, 2007). It is cited here as a map, not as a source of numbers; the reviewed recorded record has no abstract.

The helix was not inferred from a supplement label. Fibre diffraction in the 1950s produced a three-chain model. Rich and Crick published the molecular structure in 1961 (Rich and Crick, 1961). Ramachandran’s earlier diffraction papers sit in the same lineage (Ramachandran and Kartha, 1954). Bella later reviewed the peptide crystallography that filled the model with water bridges and side-chain geometry (Bella, 2016). Those papers are structure papers. They license no claim about a scoop of hydrolysate.

The repeating unit is Gly-X-Y. Glycine is required because there is no room for a side chain at the supercoil axis. Proline and hydroxyproline favour the polyproline-II geometry that makes the individual chains. The hydroxylation of proline is an enzymatic, ascorbate-dependent post-translational modification, not a dietary decoration. That last point is developed in section 02, because it is the place where a vitamin deficiency and a collagen-peptide advertisement are most often allowed to trade clothes.

02Glycine, proline, hydroxyproline, and vitamin C

The amino-acid signature of collagen is not a marketing slogan. It is why the protein can exist. Glycine occupies every third position. Proline and 4-hydroxyproline occupy much of X and Y. Hydroxyproline is not coded. It is made on the assembled chain by prolyl hydroxylase, and that enzyme requires ascorbate. Peterkofsky stated the requirement and the structural consequence: hydroxyproline stabilises the triple helix and stimulates procollagen secretion (Peterkofsky, 1991). That is established biochemistry. It is not a licence to treat a collagen-peptide drink as a vitamin-C therapy, or a vitamin-C tablet as a collagen-peptide therapy.

Scurvy is the human disease of ascorbate deficiency, and extracellular-matrix collapse is part of its picture. The mechanism is not only a missing hydroxyl. In scorbutic or fasted guinea pigs, serum suppresses chondrocyte collagen and proteoglycan synthesis even when ascorbate is supplied in the dish; humoral IGF-binding proteins travel with the weight loss (Oyamada, Bird and Peterkofsky, 1988). That paper is an animal and cell result. It is cited to stop a one-step story: no ascorbate, therefore no hydroxyl, therefore no helix, therefore a powder will repair the face. The story is too short.

Eastoe tabulated the amino-acid composition of mammalian collagen and gelatin in 1955 (Eastoe, 1955). The paper exists. The harvested record has no abstract. Residue-level percentages, and any claim that tryptophan is exactly zero grams per hundred grams, are therefore not printed from memory. What can be said without that table is still enough for later sections. Circulating peptides after hydrolysate ingestion are hydroxyproline-containing, with Pro-Hyp dominant (Iwai et al., 2005). Independent feeding trials use collagen peptides as the low-leucine, low-tryptophan comparator against dairy proteins (Oikawa et al., 2020). Shoulders and Raines already placed glycine, proline, and hydroxyproline at the centre of the helix (Shoulders and Raines, 2009). The protein is Gly/Pro/Hyp-rich. It is not, on that account, a high-quality muscle protein. Section 13 keeps those two facts from trading places.

03Collagen types

There is not one collagen. There is a family of trimeric proteins that share at least one triple-helical domain and differ in everything else that matters for a tissue. Ricard-Blum counted 28 members (Ricard-Blum, 2011). The count has been a moving integer: Prockop and Kivirikko already had at least 19 types in 1995 (Prockop and Kivirikko, 1995); Gelse, Pöschl and Aigner had more than 20 in 2003 (Gelse, Pöschl and Aigner, 2003); Myllyharju and Kivirikko had at least 27 types and 42 distinct polypeptide chains in 2004 (Myllyharju and Kivirikko, 2004). The number is established as large and still slightly date-dependent. The article does not need a final census. It needs the types that dominate the tissues people buy collagen for.

Type I is the workhorse fibril of skin, tendon, ligament, and bone. Type II is the fibril of hyaline cartilage. Type III accompanies type I in skin and hollow organs. Types V and XI nucleate those fibrils (Kadler, Hill and Canty-Laird, 2008; Shoulders and Raines, 2009). The remaining types are FACITs, multiplexins, membrane collagens, and other specialists. They are not interchangeable with a bovine hide hydrolysate. An oral peptide mix is not a tissue-targeted drug that “goes to type II cartilage” because the word collagen appears on both sides of the sentence.

Gelse, Pöschl and Aigner mapped type distribution onto function (Gelse, Pöschl and Aigner, 2003). Thorpe and Screen described tendon as a hierarchical, type-I-rich composite, not a bag of loose fibrils (Thorpe and Screen, 2016). Those reviews are the tissue grammar. They do not say that swallowing peptides rebuilds the hierarchy.

The commercial literature treats that grammar as optional. A hide hydrolysate is sold as if it were cartilage type II. A chicken sternal extract is sold as if it were dermal type I. An undenatured 40 mg capsule is sold next to a 10 g drink as if both were “collagen.” Section 04 therefore maps the five matrices onto the instruments later sections actually use, not onto the word on the tub. A reader who keeps type, tissue, and endpoint in separate columns will not need the rest of this part restated.

04Connective-tissue map: skin, cartilage, tendon, ligament, bone

Skin is a type-I/III-rich dermis under a renewing epidermis. Cartilage is a type-II-rich, avascular, load-bearing gel. Tendon and ligament are aligned type-I composites that transmit force; they are not the same organ, and they do not turn over like muscle. Bone is a mineralised type-I scaffold. The disc that sits between vertebrae is another collagen-rich tissue with its own residence time. These matrices share a triple helix and almost nothing else that an outcome trial can treat as one endpoint.

That map is established as anatomy. It is also the first filter on the commercial literature. A wrinkle-volume trial is a skin-instrument trial. A WOMAC trial is a symptom trial in a joint. A VISA-A trial is a tendinopathy questionnaire. A T-score trial is a densitometry trial. A fat-free-mass trial is a composition-machine trial. None of those instruments is “connective tissue.” The word is a category, not a measurement.

05Turnover

Collagen is not a static rope, and it is not a rapidly turning muscle protein. The residence time depends on the tissue and on whether the tissue is healthy. That distinction is established and is the hardest fact in this part for a daily-powder story to survive.

Heinemeier and colleagues measured bomb-¹⁴C in forensic Achilles cores. The carbon in the healthy core tracked atmospheric ¹⁴C from the first seventeen years of life. Muscle from the same bodies turned over continuously (Heinemeier et al., 2013). A later paper from the same group showed that tendinopathic matrix has years of abnormal turnover before pain (Heinemeier et al., 2018). Pathology is not the healthy core. A loading-plus-peptide trial in students is not a refutation of the bomb-pulse stillness of an adult Achilles interior.

Cartilage collagen is not young either. Verzijl and colleagues used accumulated advanced glycation end-products and the racemisation of aspartate to show that cartilage collagen residence time is measured in decades, and is slower than skin (Verzijl et al., 2000). Sivan and colleagues estimated normal disc collagen half-lives on the order of a century: a turnover constant of about 0.00728 per year between ages 20 and 40 (half-life 95 years) and 0.00323 per year between 50 and 80 (half-life 215 years) (Sivan et al., 2008). Degenerate discs may turn over faster; the statistics were uncertain in that paper. The numbers are cited as published estimates, not as treatment clocks.

The implication for later sections is narrow and unkind to slogans. If the core of a healthy tendon is not renewed after growth, an oral peptide cannot be sold as a general tendon-rebuilding protein. If cartilage collagen lives for decades, a twelve-week visual-analogue change is a symptom change until imaging or structure says otherwise. If skin collagen turns over faster than cartilage, that still does not make a corneometer increment into a new dermis. Turnover sets the prior. Trials still have to earn their endpoints.

06How this article grades evidence

The grades are used in the sentence that carries the claim.

Established means a structural, biochemical, or demographic fact that is not in serious dispute among the papers that measured it: the triple helix, the ascorbate requirement for prolyl hydroxylation, the existence of many collagen types, the bomb-pulse stillness of healthy Achilles core, the FAO definition of protein quality.

Strongly supported means a human finding that has been measured more than once, or measured once with a method that leaves little room for the opposite reading: Pro-Hyp appearance after hydrolysate ingestion; whey exceeding collagen for myofibrillar synthesis; hydrolysate batches differing in peptide maps.

Emerging means a statistically detectable human signal that remains product-specific, short, sponsor-adjacent, or instrument-bound: VERISOL wrinkle volume, TENDOFORTE VISA-A in a twenty-person crossover, König T-score deltas, most skin meta-analyses.

Speculative means a class-wide leap: any collagen powder rebuilds skin, heals tendons, prevents fractures, or builds muscle as a complete protein would.

Animal and cell findings identify a peptide, an enzyme, or a circulating species. They are never phrased as human outcomes. Conflict is left in conflict. A newer null does not automatically retire an older positive, and twenty papers from one industrial peptide platform are not twenty independent discoveries of collagen. Amounts are trial facts. They are not recommendations.

07Gelatin is not a collagen peptide

Gelatin is collagen that has been heated until the triple helix comes apart and the chains remain long. It gels. It is a food colloid. It is not a defined oligopeptide mix. Van Vijven, Luijsterburg, Verhagen and colleagues already separated undenatured collagen, gelatin, and collagen hydrolysate when they reviewed osteoarthritis supplements (Van Vijven et al., 2012). Martínez-Puig, Costa-Larrión and colleagues restated the chemical split: native collagen is an epitope story; hydrolysates are a peptide story (Martínez-Puig et al., 2023). That distinction is established. Marketing collapses it.

The Baar-laboratory loading studies administer gelatin, not a patented bioactive peptide brand. Shaw, Lee, Aidoo and Baar gave vitamin-C-enriched gelatin before rope-skipping in eight men and measured a collagen-synthesis marker and an engineered-ligament outcome (Shaw et al., 2017). Lis and Baar later compared 15 g of vitamin-C gelatin, hydrolyzed collagen, a gummy, and placebo in ten men; PINP tended to rise about 20% with gelatin or hydrolysate and did not with the gummy or placebo, while amino acids rose in all protein arms (Lis and Baar, 2019). Those papers are gelatin-plus-loading papers. They are cited in Part Three as biomarker work. They are not VERISOL papers, and they are not a licence to treat culinary gelatin, a grocery hydrolysate, and a patented fraction as one scoop.

08Hydrolysis and molecular-weight distributions

Hydrolysis cuts gelatin into peptides. The cut is enzymatic or acid, the resulting map is a distribution, and the distribution is not one number. “Low molecular weight” is a marketing and analytical phrase, not a class constant. Kim’s LMWCP was specified as a fish hydrolysate with Gly-X-Y tripeptide content greater than 15%, including 3% Gly-Pro-Hyp (Kim et al., 2018). Schauss, Stenehjem and colleagues described BioCell as a low-molecular-weight chicken sternal extract (Schauss et al., 2012). Those are product descriptions. No harvested paper gives a single molecular-weight that all collagen peptides share. The grade for a universal megadalton claim is speculative.

What hydrolysis does do is make absorption a different experiment from eating a steak. Oesser, Adam, Babel and Seifert fed ¹⁴C-labelled gelatin hydrolysate to mice: 95% of the label was absorbed within twelve hours, cartilage radioactivity exceeded a proline control by more than twofold, and peptides of 2.5–15 kD survived gut passage (Oesser et al., 1999). That is a mouse and gut-sac result. It is strongly supported as absorption physiology. It is an inference, not a finding, that human joints repair themselves because a mouse cartilage took up label.

In fasted volunteers, 9.4–23 g of porcine or chicken gelatin hydrolysate produced peptide-form hydroxyproline of 20–60 nmol/mL at one to two hours, with Pro-Hyp the dominant circulating peptide and Ala-Hyp and Pro-Hyp-Gly among the minors (Iwai et al., 2005). Yazaki, Hashimoto and colleagues later identified 17 peptide species in human plasma after a high-tripeptide hydrolysate; Gly-Pro-Hyp was enriched in blood, and mouse skin enriched Pro-Hyp after Gly-Pro-Hyp hydrolysis (Yazaki et al., 2017). Appearance is established. “Therefore the dermis remakes itself” is not.

The batches are not interchangeable even at this level. Schadow, Siebert and colleagues, and later Simons, Schadow and colleagues, showed that bovine hydrolysate batches differ in peptide composition and in how they move human osteoarthritic cartilage and synoviocytes; some Peptan-class materials appear in that work (Schadow et al., 2013; Schadow et al., 2017; Simons et al., 2018). Inoue, Sugihara and Wang showed that two hydrolysates with different Pro-Hyp and Hyp-Gly contents are not equal on skin instruments (Inoue et al., 2016). Those papers are strongly supported as a warning against “any collagen powder.”

09Source species

Trials use bovine hide, porcine skin, chicken sternal cartilage, marine and freshwater fish, and mixed sources. Iwai’s circulating-peptide work used porcine and chicken hydrolysates (Iwai et al., 2005). Kim’s LMWCP was fish (Kim et al., 2018). Evans used a freshwater marine hydrolysate sold as Vinh Wellness Collagen (Evans et al., 2021). Schauss and the UC-II papers use chicken sternal material, one hydrolyzed and one undenatured (Schauss et al., 2012; Lugo et al., 2016). Species is established as a label variable. It is not shown, in the reviewed record, to be the active variable once peptide map, dose, and co-ingredients differ. A marine claim is not automatically a better claim. A bovine claim is not automatically a worse one. The swallowed composition is the thing that was tested.

10Branded bioactive collagen peptides

GELITA’s published family — VERISOL, FORTIGEL, FORTIBONE, BODYBALANCE, TENDOFORTE — is a set of specified bioactive collagen peptide compositions, not a synonym for collagen. The product literature is used in this article only to identify which mix a trial administered. It is not evidence of effect. The trials that name those mixes, or that write “specific collagen peptides” with the Oesser–König–Zdzieblik–Jerger author set, occupy most of the positive skin, joint, tendon-morphology, bone-density, and fat-free-mass file. Part Four keeps a ledger. The point here is identity. A result attaches to the fraction that was swallowed.

Proksch, Schunck, Zague and colleagues named VERISOL in the eye-wrinkle trial: 2.5 g/day for eight weeks in 114 women aged 45–65 years (Proksch et al., 2014b). Praet, Purdam, Welvaert and colleagues named TENDOFORTE in the Achilles crossover (Praet et al., 2019). König, Oesser, Scharla and Zdzieblik’s twelve-month T-score trial is the FORTIBONE-class bone paper in this file (König et al., 2018). Zdzieblik, Oesser, Baumstark and colleagues’ sarcopenia trial, and the later Jendricke, Oertzen-Hagemann and Kirmse resistance-training papers, are the BODYBALANCE-era fat-free-mass cluster (Zdzieblik et al., 2015; Jendricke et al., 2019; Oertzen-Hagemann et al., 2019; Kirmse et al., 2019). Those names are not decorations. They are the independent variable.

11Other branded hydrolysates

Not every named product is a GELITA fraction. Asserin, Lati, Shioya and Prawitt sit in the Peptan-class orbit (Asserin et al., 2015). Clark, Sebastianelli, Flechsenhar and colleagues administered CH-Alpha, 10 g/day as a liquid, in athletes without diagnosed osteoarthritis (Clark et al., 2008). Schauss, Stenehjem and colleagues administered BioCell Collagen, 2 g/day of a chicken sternal extract, for 70 days in hip or knee osteoarthritis (Schauss et al., 2012). Kim’s LMWCP and Evans’s Vinh Wellness Collagen are defined marine or fish hydrolysates (Kim et al., 2018; Evans et al., 2021). Bolke’s ELASTEN is a 2.5 g peptide ampoule plus vitamin C, zinc, biotin, and vitamin E (Bolke et al., 2019). Elam, Elam and colleagues’ bone trial is a calcium–collagen chelate against calcium plus vitamin D, not peptides alone (Elam et al., 2015). None of these generalises to an unassayed grocery tub. The grade for that negative generalisation is strongly supported.

12Undenatured type II is a different product

UC-II is undenatured type II collagen, typically administered at 40 mg/day. The hypothesized mechanism is oral tolerance to native epitopes, not delivery of grams of type I oligopeptides. Martínez-Puig and colleagues state that chemical split (Martínez-Puig et al., 2023). Crowley, Lau, Klimenko and colleagues compared UC-II with glucosamine plus chondroitin for 90 days without a placebo (Crowley et al., 2009). Lugo, Saiyed, Lau and colleagues later ran a 40 mg stepmill trial in 55 healthy adults (Lugo et al., 2013) and a 191-person, 180-day osteoarthritis trial against placebo and against glucosamine plus chondroitin (Lugo et al., 2016). Those papers are kept in the joint table so a reader can see them. They are not collagen-peptide results. Pooling them with VERISOL or FORTIGEL is a category error. The grade for the identity claim is established.

Figure 4 is the identity diagram for the rest of the document. Five objects share an aisle word. Only two of them are collagen peptides in the sense this article tests, and those two still split into generic hydrolysate and branded bioactive mix. The next section is protein quality, which is a different test and a different failure. A reader who keeps the five objects in separate columns will not need Part Four to invent a sixth.

13Protein quality: why collagen fails as a complete protein

Protein quality is the ability of a food to supply the nine dietary essential amino acids relative to requirement, corrected for digestibility. Wolfe, Church, Ferrando and Moughan restated that definition and the current score, DIAAS (Wolfe et al., 2024). The FAO Expert Consultation that recommended DIAAS in place of PDCAAS treats amino acids as individual nutrients, uses the lowest digestible indispensable amino-acid ratio, and includes tryptophan in the scoring pattern (FAO, 2013). A protein that is nearly devoid of any one indispensable amino acid cannot be a complete protein, however digestible the rest of the chain is. That framework is established.

Collagen is not in Gorissen, Crombag, Senden and colleagues’ UPLC panel of isolates (Gorissen et al., 2018). That paper is cited for the proteins that were assayed: whey 43% essential amino acids, milk 39%, human muscle 38%, casein 34%, egg 32%, wheat/lupin/oat about 21–22%. No invented collagen percentage is placed beside that table. The cleanest harvest evidence that collagen-peptide supplements, as administered, do not match dairy proteins for the amino acids that gate muscle protein synthesis is Oikawa, MacDonald, Ogborn and colleagues’ comparison with α-lactalbumin: the dairy protein raised plasma leucine and tryptophan relative to collagen (P < 0.001 for both) (Oikawa et al., 2020b). Combined with the helix composition (Shoulders and Raines, 2009) and the hydroxyproline-rich circulating peptides (Iwai et al., 2005), the incompleteness claim is strongly supported. A numeric DIAAS for collagen, and a printed “zero grams of tryptophan,” are not extracted from Eastoe 1955 in this build and are not invented.

Physiology then does what the score predicts. Phillips and Van Loon placed leucine and the other branched-chain amino acids at the centre of the post-exercise synthetic response and discussed athlete protein in the 1.3–1.8 g·kg⁻¹·d⁻¹ range as a consensus frame, not as a recommendation in this document (Phillips and Van Loon, 2011). Churchward-Venne, Burd, Clements and colleagues showed that 25 g of whey kept myofibrillar synthesis elevated three to five hours after exercise, and that 6.25 g of whey plus leucine, or plus essential amino acids without matching leucine, did not (Churchward-Venne et al., 2012). Oikawa, Kamal, Saddler and colleagues then put collagen in the same room as whey: 22 older women, about 69 years, 30 g of whey or 30 g of collagen peptides twice daily for six days, unilateral resistance exercise. Acute myofibrillar synthesis on whey was 0.017%/h at rest and 0.032%/h after exercise; collagen reached only 0.012%/h after exercise; whey exceeded collagen in both legs (P = 0.02); longer-term synthesis rose with whey and not significantly with collagen (Oikawa et al., 2020a). Jacinto, Nunes, Gorissen and colleagues leucine-matched the comparison over ten weeks of training: 35 g whey (3.0 g leucine) versus 35 g collagen peptides (1.0 g leucine plus 2.0 g free leucine). Vastus lateralis thickness rose 8.4% on whey and 5.6% on collagen (Jacinto et al., 2022). Free leucine did not convert collagen into whey. Robberechts, Delecluse and colleagues replaced part of a 45 g whey allotment with 20 g of collagen peptides during eccentric training and found no recovery advantage (Robberechts et al., 2024). Kirmse, Wackenhage and colleagues gave 15 g of hydrolyzed collagen twice daily during one week of heavy resistance training and did not raise muscle connective-protein synthesis versus placebo (Kirmse et al., 2024).

The NIH Office of Dietary Supplements fact sheet on exercise and athletic performance states a 1.2–2.0 g/kg protein range for athletes and does not list collagen among ingredients with performance evidence. That sheet is a labelled secondary instrument, not a trial. It is consistent with the tracer file. Collagen peptides are a specialty glycine/proline/hydroxyproline delivery. They are not a whey substitute. Any later paragraph that reports extra fat-free mass on a composition machine without mentioning Oikawa and Jacinto is incomplete.

The helix is not on trial. The powder is. The trials that follow measure hydration, elasticity, wrinkle volume, joint pain, tendon size, bone density, and fat-free mass. Almost all of them administer a named peptide mix, a manufacturer-class hydrolysate, or a blend. The endpoints are devices and questionnaires. The correct grade, across this part, is almost never higher than emerging.

14Skin hydration

Hydration is the most repeated skin instrument in this file and the easiest to over-read. Corneometry reports stratum-corneum water. It does not report dermal collagen mass, photoageing grade, or a clinician’s decision that a face has been treated. The honest grade for a hydration signal from named hydrolysates is emerging. The grade for a class-wide grocery-powder claim is speculative.

A pair of randomised, placebo-controlled trials summarised by Asserin, Lati, Shioya and Prawitt reported that oral collagen peptides increased skin hydration at eight weeks and raised dermal ultrasound density while reducing a fragmentation index (Asserin et al., 2015). The abstract does not give a complete analysed n. The work sits in the Peptan-class orbit. Those two facts travel with the result. They do not cancel a corneometry difference. They stop anyone from treating the paper as a test of an uncharacterised tub.

A double-blind, randomised, placebo-controlled trial of a defined low-molecular-weight collagen peptide assigned 64 participants to 1,000 mg/day of LMWCP or placebo for twelve weeks (Kim et al., 2018). The product was specified as a fish hydrolysate with Gly-X-Y tripeptide content greater than 15%, including 3% Gly-Pro-Hyp. Hydration was higher versus placebo at six and twelve weeks. The authors recorded no adverse symptoms attributed to the test material. The trial is small, product-specific, and twelve weeks long. It is a hydration paper for that peptide map. It is not a hydration paper for collagen as a word.

The earlier specific-hydrolysate trial by Proksch, Segger, Degwert and Schunck is often cited for moisture. It should not be. In that three-arm, double-blind, placebo-controlled study, 69 women aged 35–55 years received 2.5 g or 5.0 g of a specific collagen hydrolysate, or placebo, once daily for eight weeks (Proksch et al., 2014a). Elasticity was the primary interest and moved versus placebo at both reported amounts. Moisture and transepidermal water loss improved only in a subgroup analysis and failed to reach significance in the whole sample. A moisture claim built on that paper is a subgroup claim.

Inoue, Sugihara and Wang ran a three-arm, randomised, double-blind, placebo-controlled comparison of two hydrolysates that differed in Pro-Hyp and Hyp-Gly content (Inoue et al., 2016). The higher-dipeptide preparation produced larger improvements in moisture, elasticity, wrinkles, and roughness than the lower-dipeptide hydrolysate or placebo over eight weeks. Safety was checked by blood test. The analysed n is not fully numeric in the reviewed recorded abstract. The compositional contrast is the load-bearing finding: two products that can both be sold as collagen hydrolysate are not interchangeable on the instruments this section uses.

Bolke, Schlippe, Gerß and Voss randomised 72 women aged 35 years or older to a twelve-week drinkable ampoule or placebo (Bolke et al., 2019). Hydration by corneometry improved versus placebo, as did elasticity, roughness, and ultrasound density, and the differences were retained in a four-week off-product follow-up. The ampoule was ELASTEN^®. It delivered 2.5 g of collagen peptides together with acerola extract, vitamin C, zinc, biotin, and a native vitamin E complex. The trial is a blend trial. Attribution of the hydration change to collagen peptides alone is not licensed.

Three systematic reviews then pool these products. Choi, Sung, Juhasz and Mesinkovska reviewed 11 randomised, placebo-controlled dermatology trials in 805 subjects, most using 2.5–10 g/day of hydrolysate for 8–24 weeks, and called the signal preliminary (Choi et al., 2019). de Miranda, Weimer and Rossi meta-analysed 19 double-blind controlled trials in 1,125 participants, 95% of them women, aged 20–70 years, and reported a grouped advantage for hydrolyzed collagen on hydration, elasticity, wrinkles, and firmness (de Miranda et al., 2021). Pu, Huang, Pu and Kang pooled 26 randomised trials in 1,721 subjects and obtained a hydration Z of 4.94 and an elasticity Z of 4.49 versus placebo; source species and duration modified the hydration result, and the authors noted bias in the included trials (Pu et al., 2023). The reviews are composition-blind. They measure the literature that exists. They do not test a store-brand powder that was never in any of the trials.

The pharmacokinetic background is established and is not a hydration outcome. After 9.4–23 g of porcine or chicken gelatin hydrolysate in fasted volunteers, peptide-form hydroxyproline rose to 20–60 nmol/mL at one to two hours, with Pro-Hyp the dominant circulating species (Iwai et al., 2005). Yazaki and colleagues later identified 17 collagen-derived peptide species in human plasma after a high-tripeptide hydrolysate, with Gly-Pro-Hyp enriched in blood and Pro-Hyp enriched in mouse skin after Gly-Pro-Hyp hydrolysis (Yazaki et al., 2017). Human dermal enrichment was not measured. Culture papers at 200 nmol/mL Pro-Hyp report fibroblast outgrowth, a 1.5-fold rise in human dermal-fibroblast proliferation, a 3.8-fold rise in hyaluronan, and a 2.3-fold rise in HAS2 transcript (Shigemura et al., 2009; Ohara et al., 2010). Those concentrations are culture conditions. They are not proven human-dermis pharmacology, and they do not convert a corneometer increment into a disease outcome.

15Elasticity

Cutometry is a suction measurement. It is more specific than hydration and still not a clinical success criterion. The grade for an elasticity signal from named bioactive mixes is emerging.

The load-bearing elasticity trial is the 69-woman study already introduced (Proksch et al., 2014a). Both the 2.5 g/day and the 5.0 g/day arms of the specific hydrolysate improved elasticity versus placebo after eight weeks. In a four-week regression phase, a higher elasticity level persisted in the older women. The paper and the later VERISOL wrinkle trial share an author set. Treating them as two independent discoveries of “collagen” is a category error. Treating them as two measurements of one manufacturer-adjacent peptide family is accurate.

Kim’s LMWCP trial moved two of three elasticity parameters versus placebo at twelve weeks in 64 participants on 1,000 mg/day (Kim et al., 2018). One of the three parameters improved from baseline in the LMWCP arm. That is a partial instrumental win on a defined tripeptide mix, not a full biomechanical reconstruction of dermis.

Evans, Lewis, Zakaria and Pelipyagina ran a triple-blind, placebo-controlled, parallel trial of a freshwater marine hydrolysate (Vinh Wellness Collagen) in women aged 45–60 years, with assessments at 0, 6, and 12 weeks by VISIA, Cutometer, and a self-score (Evans et al., 2021). Elasticity improved in a planned 45–54-year subgroup: cheek elasticity rose 20% from baseline at six weeks and 10% at twelve weeks. The whole-sample elasticity claim is weaker than the subgroup. A one-sided facial contrast (a 24% greater wrinkle reduction on the right side versus placebo) is a weak way to present a bilateral organ.

Pu’s meta-analysis reported an elasticity Z of 4.49 versus placebo across 26 trials and found no significant difference by source species or measurement method on that endpoint (Pu et al., 2023). The pooled Z is a statement about a heterogeneous, sponsor-heavy, mostly female file. It is not a statement that elasticity will rise in a man who buys an unassayed powder.

What elasticity trials do not show belongs in the same paragraph. They do not show restoration of solar elastosis. They do not show a dermatologist global that would satisfy a drug label. They do not show that the cutometer change survives a year off product. They do not show that a gelatin dessert, a generic hydrolysate, and VERISOL are the same intervention. Appearance of Pro-Hyp in blood after gelatin hydrolysate (Iwai et al., 2005) is common to many oral collagen products. Equality of oligopeptide maps is not (Inoue et al., 2016).

Section 16 then asks the wrinkle question on the same products, because a reader who has just seen a cutometer win will be offered a PRIMOS win as if it were a clinical treatment. It is not. The devices remain devices. The mixes remain named. Figure 6 collects the three skin instruments on one plate so the later wrinkle paragraph does not have to redraw the file. A meta-analysis that pools those plates is a review of named mixes, not a test of grocery collagen.

16Wrinkles

Wrinkle volume is the skin endpoint that sells the tub. It is also the endpoint that is most clearly attached to a trademark. The grade for a named bioactive collagen peptide reducing eye-wrinkle volume versus placebo is emerging. The grade for a clinically meaningful, class-wide anti-wrinkle treatment is speculative.

A double-blind, placebo-controlled trial randomised 114 women aged 45–65 years to 2.5 g/day of the specific bioactive collagen peptide VERISOL or placebo for eight weeks, 57 per arm (Proksch, Schunck, Zague et al., 2014b). Eye-wrinkle volume fell versus placebo at four and eight weeks, quoted as a 20% reduction at eight weeks, and the difference persisted in a four-week off-product regression phase. A suction-blister subgroup showed increases in procollagen I, elastin, and fibrillin after eight weeks. The trial is the strongest single wrinkle paper in the reviewed record. It is a VERISOL paper. The abstract names the product. The duration is eight weeks plus four weeks of follow-up. The endpoint is instrumental wrinkle volume, not a five-year photoageing hard outcome.

Kim’s LMWCP trial improved a visual wrinkle score and three wrinkle parameters versus placebo at twelve weeks on 1,000 mg/day in 64 participants (Kim et al., 2018). Evans’s marine hydrolysate reduced a wrinkle score 35% from baseline at twelve weeks (P = 0.035) in women aged 45–60 years (Evans et al., 2021). Baseline-to-end change is not the same statistical object as a placebo-adjusted difference. The right-side contrast noted above remains a presentation weakness. Inoue’s higher-dipeptide hydrolysate again outperformed the lower-dipeptide product on wrinkles (Inoue et al., 2016). Composition dependence is the through-line, not a species romance.

An open-label VERISOL series in 25 participants reported 12% faster nail growth and 42% fewer broken nails over 24 weeks at 2.5 g/day (Hexsel et al., 2017). It is named, uncontrolled, and not a wrinkle RCT. It is listed so that a reader who meets it in a brochure knows what it is.

de Miranda and Pu both pool wrinkle or related ageing instruments into a favourable hydrolyzed-collagen estimate (de Miranda et al., 2021; Pu et al., 2023). The objection that those reviews are large is true and incomplete. They are large because they add branded mixes, marine hydrolysates, and at least one multi-ingredient blend. They are short because almost every trial is eight to twelve weeks. They are female. They are device-scaled. No harvest RCT used a dermatologist global clinical-success criterion as the sole primary that would satisfy a medicinal-product label (Proksch et al., 2014b; Bolke et al., 2019; Evans et al., 2021). That is not a quibble about taste. It is the difference between a millimetre-scale instrument and a treatment claim.

Amounts in this part are the amounts the cited trials administered, always with the population and the duration. They are not instructions for use.

17Joint pain and osteoarthritis

The joint file is older, larger, and more internally contradictory than the skin file. Some visual-analogue scores move. WOMAC pain often does not. A multinational pharmaceutical-grade hydrolysate trial was null in the total sample and positive in German sites and in more-severe patients. Undenatured type II collagen is a different molecule and is kept in a separate box. The grade for a modest symptom signal from particular hydrolysates is emerging. The grade for disease modification is unmeasured. The Moskowitz multinational result is strongly supported as a qualified null.

Clark, Sebastianelli, Flechsenhar and Aukermann ran a 24-week, randomised, placebo-controlled, double-blind trial of liquid CH-Alpha at Penn State (Clark et al., 2008). One hundred forty-seven varsity or club athletes with activity-related joint pain and no diagnosed joint disease were assigned to 25 mL containing 10 g collagen hydrolysate or to a xanthan placebo; 97 of 147 were evaluable. Visual-analogue scales for pain, mobility, and inflammation were the reported endpoints. Attrition of that size is a finding, not a footnote. The population is not radiographic osteoarthritis.

Benito-Ruiz and colleagues randomised 250 subjects with primary knee osteoarthritis to 10 g/day collagen hydrolysate for six months in a multicentre, double-blind trial framed as a food-ingredient study (Benito-Ruiz et al., 2009). Knee comfort improved on visual-analogue pain and on the WOMAC pain subscale. The authors emphasised greater benefit in subjects with more deteriorated joints and in those whose habitual diet was low in meat protein. Subgroup emphasis after a comfort win is how a modest overall signal is made to look decisive.

Bruyère, Zegels, Leonori and Rabenda randomised 200 men and women aged 50 years or older with joint pain of at least 30 mm on a visual-analogue scale to 1,200 mg/day collagen hydrolysate or placebo for six months (Bruyère et al., 2012). The primary analysis was the proportion of responders, defined as a 20% or greater improvement in the most painful joint. At six months the responder rate was 51.6% versus 36.5% (P < 0.05). At three months it was 44.1% versus 39.6% (P = 0.53). The absolute difference at six months is 15 percentage points. Separation was delayed. Security and tolerability did not differ between arms.

A 24-week pilot randomised, placebo-controlled, double-blind trial in 30 people with mild knee osteoarthritis used delayed gadolinium-enhanced magnetic resonance imaging of cartilage as the primary instrument (McAlindon et al., 2011). Medial and lateral tibial dGEMRIC T1 rose by a median 29 and 41 ms on collagen hydrolysate and fell by 37 and 36 ms on placebo at 24 weeks. Participants were allowed to continue prior analgesics. Symptoms were not the win. n is 30. The imaging contrast is real in the sample and is not structural disease modification by any regulatory standard.

The paper that should sit next to every positive hydrolysate brochure is Moskowitz’s review of the pharmaceutical-grade collagen hydrolysate programme (Moskowitz, 2000). In a multicentre, randomised, double-blind, placebo-controlled trial conducted in the United States, the United Kingdom, and Germany, mean pain scores did not differ for the total study group. German sites showed a treatment advantage. Patients with more severe symptoms at entry showed increased efficacy versus placebo. Clinical use was associated with minimal adverse effects, mainly gastrointestinal fullness or unpleasant taste. Site-and-severity slicing of a null total sample is the oldest move in this literature. Bello and Oesser’s narrative review of hydrolysate in osteoarthritis inherits the same industrial adjacency (Bello and Oesser, 2006).

Zdzieblik, Oesser, Gollhofer and König randomised 139 athletic subjects with functional knee pain to 5 g/day bioactive collagen peptides or placebo for twelve weeks (Zdzieblik et al., 2017). Activity-related visual-analogue change was 19.5 versus 13.9 (P = 0.046). Physician-rated activity pain also separated. Rest pain did not (P = 0.209). Placebo moved. The between-group remainder is small. The author set is the Gelita-orbit cluster that reappears in the bone and fat-free-mass papers.

Schauss, Stenehjem, Park and Endres randomised 80 people with physician-verified progressive hip or knee osteoarthritis and baseline pain of 4 or higher to 2 g/day BioCell Collagen, a low-molecular-weight hydrolyzed chicken sternal cartilage extract, or placebo for 70 days (Schauss et al., 2012). Intent-to-treat visual-analogue pain was lower on day 70 (P < 0.001), and WOMAC scores moved with it. Tolerability was comparable to placebo. BioCell is a named chicken-sternal extract, not a bovine type-I peptide mix and not undenatured type II.

The meta-analyses refuse a single OA effect size. Van Vijven, Luijsterburg, Verhagen and van Osch already separated undenatured collagen, gelatin, and collagen hydrolysate (Van Vijven et al., 2012). Eight studies met their criteria. The pooled WOMAC-pain mean difference for hydrolysate versus placebo, from three studies, was −0.49 (95% CI −1.10 to 0.12), not significant. Some visual-analogue or glucosamine-comparison contrasts were positive. García-Coronado and colleagues later reported a WOMAC-total weighted mean difference of −8.00 (95% CI −13.04 to −2.95) favouring collagen supplements, with stiffness significant and pain not significant in the quoted subgroup (García-Coronado et al., 2019). Honvo, Lengelé, Charles and Reginster mapped 25 clinical papers on hydrolysate or undenatured collagen and found no disease-modifying osteoarthritis drug (Honvo et al., 2020). Martínez-Puig, Costa-Larrión, Rubio-Rodríguez and Gálvez-Martín restated the chemical split: native collagen is hypothesised to act by epitope recognition; hydrolysates, if they act, act as peptides (Martínez-Puig et al., 2023). A 2025 systematic review of type-I hydrolysate reported joint outcomes more often than bone conclusions (Brueckheimer et al., 2025). Instrument-dependent, composition-mixed, and DMOAD-negative is the accurate review-level sentence.

Oesser, Adam, Babel and Seifert showed that 95% of enterally applied ¹⁴C-gelatin hydrolysate was absorbed within twelve hours in mice and that cartilage radioactivity exceeded a proline control by more than twofold, with peptides of 2.5–15 kD detected after gut passage (Oesser et al., 1999). Absorption and murine cartilage label are strongly supported. Human joint repair is an inference from that paper, and a weak one. Schadow 2013, Schadow 2017, and Simons 2018 then showed that hydrolysate batches differ in peptide composition and in how they move human osteoarthritic cartilage and synoviocytes (Schadow et al., 2013; Schadow et al., 2017; Simons et al., 2018). A powder without a published peptide map is not the powder in Clark, Benito-Ruiz, or Zdzieblik 2017.

Section 18 therefore does not inherit a repaired joint. It asks whether a named peptide plus loading can change a tendon questionnaire or a tendon image, and it keeps Heinemeier’s still Achilles core in the same paragraph as Praet and Jerger. A reader who wants a disease-modifying osteoarthritis claim will not find one by turning the page.

18Tendon adaptation and rehabilitation adjuncts

Healthy adult Achilles core collagen is essentially not renewed after growth. Heinemeier, Schjerling, Heinemeier and Magnusson measured bomb-¹⁴C in 28 forensic Achilles cores and four skeletal-muscle samples (Heinemeier et al., 2013). Tendon ¹⁴C tracked atmospheric levels from the first approximately 17 years of life. Muscle turned over continuously. Tendinopathic matrix is a different tissue: bomb-pulse dating of 25 tendinopathic and 10 healthy Achilles tendons showed that about half the tendinopathic collagen had undergone years of abnormal turnover before symptoms (Heinemeier et al., 2018). Pathology is not healthy-core biology. Oral peptides are not a tissue-targeted drug. Thorpe and Screen’s structural review remains the map of a hierarchical, type-I-rich tendon, not a rationale for a scoop (Thorpe and Screen, 2016).

The grade for a biomarker rise after vitamin C–enriched gelatin plus jumping is emerging. The grade for independently replicated tendinopathy care is unmeasured. Hijikata’s Achilles hydrolysate trial was not recovered by title or author harvest and is not cited.

Shaw, Lee-Barthel, Ross and Wang ran a randomised, double-blind, crossover protocol in eight healthy men (Shaw et al., 2017). Participants consumed 5 g or 15 g of vitamin C–enriched gelatin, or placebo, then completed six minutes of rope-skipping; the pattern was repeated three times daily for three days. Circulating glycine rose with the gelatin amount. Amino-terminal propeptide of type I collagen and engineered-ligament outcomes were the reported collagen-synthesis readouts. n is 8. The intervention is gelatin plus vitamin C plus loading, not hydrolyzed peptides alone, and not injury prevention.

Lis and Baar repeated the logic in ten recreationally active men, again as a randomised, double-blind crossover, comparing placebo with 15 g of vitamin C–enriched gelatin, 15 g of vitamin C–enriched hydrolysate, or a gummy containing equal parts of both, taken one hour before six minutes of jump-rope (Lis and Baar, 2019). Amino acids rose similarly across active arms. N-terminal propeptide of procollagen tended to rise about 20% from baseline after gelatin or hydrolysate and did not after placebo or the gummy. “Tended” is the authors’ word. The trial does not show tendon cross-sectional area, stiffness under load, or pain.

Praet, Purdam, Welvaert and Vlahovich conducted the only harvested clinical tendinopathy trial (Praet et al., 2019). Twenty patients with chronic mid-portion Achilles tendinopathy completed a six-month crossover in which both periods included a twice-daily calf-strengthening programme. Group AB received named TENDOFORTE specific collagen peptides for the first three months, then placebo; group BA received the reverse. After three months, VISA-A rose 12.6 points (9.7 to 15.5) in AB and 5.3 points (2.3 to 8.3) in BA. After crossover, AB rose a further 5.9 and BA rose 17.7. No adverse events were reported. Microvascularity fell in both groups and was moderately associated with VISA-A. Exercise is in both arms. The product is named. The design is a pilot. An independent group has not repeated the VISA-A result in patients.

Dressler, Gehring, Zdzieblik and Oesser randomised 50 athletes with chronic ankle instability to 5 g/day specific collagen peptides or 5 g maltodextrin for six months, with a three-month follow-up (Dressler et al., 2018). Cumberland Ankle Instability Tool and Foot and Ankle Ability Measure scores improved versus placebo. Ankle-arthrometer stiffness did not. Subjective win, mechanical null.

Jerger, Centner, Lauber and Seynnes then asked a different question in healthy men. Forty men (mean age 26.3 years) completed 14 weeks of high-load resistance training; one arm received 5 g/day specific collagen peptides and the other 5 g placebo (Jerger et al., 2022). Achilles cross-sectional area rose 11.0% versus 4.7% (P = 0.002). Plantar-flexor thickness rose 7.3% versus 2.7% (P = 0.014). Stiffness and strength rose in both arms without a between-group difference. A second randomised, placebo-controlled trial in 50 moderately active men used the same 5 g/day specific-peptide amount and the same 14-week, three-sessions-per-week programme at 70–85% of one-repetition maximum (Jerger et al., 2023). Patellar-tendon cross-sectional area increased more at 60% and 70% of length from the proximal insertion. Stiffness, rectus-femoris area, and strength again rose in both groups without a between-group difference. That is a same-laboratory morphological echo, not a Praet replication, and not a patient-care result.

Miyamoto, Ishihara, Oshima and Kawai randomised 50 sedentary young men to 10 g/day collagen peptides or placebo for 16 weeks without a stated loading programme (Miyamoto et al., 2025). Medial-gastrocnemius and Achilles stiffness and normalised rate of torque development rose in the peptide arm. Cross-sectional area and maximal voluntary isometric contraction did not. The endpoint set is different from Jerger and from Praet. A stiffness/RTD win with a cross-sectional-area null is not a confirmation of either paper.

Figure 7 keeps grams of hydrolysate, milligrams of UC-II, and the unreplicated tendon-care row on one plate. Bone is a different instrument and a different claim, and it begins in the next section without inheriting a rebuilt tendon.

Kirmse, Lottmann, Volk and de Marées assigned 25 young men to 15 g hydrolyzed collagen peptides or a non-caloric placebo twice daily during one week of strenuous resistance training, with deuterated-water labelling of daily myofibrillar and muscle connective-protein synthesis (Kirmse et al., 2024). Hydrolyzed collagen did not further increase either fractional synthesis rate. Post-absorptive glycine, proline, and hydroxyproline rose, so the product was absorbed. The “collagen feeds the tendon during training” sentence fails that tracer test. The result is strongly supported as a negative for that claim.

Khatri, Naughton, Clifford and Harper reviewed 15 randomised trials of collagen peptides plus exercise and concluded that joint function and pain were the most consistent benefits, with mixed body-composition and recovery findings and elevated collagen-synthesis markers in some designs (Khatri et al., 2021). Bischof, Moitzi, Stafilidis and König later meta-analysed longer training-plus-peptide trials and wrote the sentence the file deserves: promising results in some but not all studies (Bischof et al., 2024). Some-but-not-all is not a rehabilitation protocol.

19Bone

No paper in the reviewed record is a fracture-endpoint randomised trial. The grade for a small T-score movement on a specific collagen peptide in postmenopausal women is emerging. The grade for fracture prevention is unmeasured. Converting a T-score delta into a broken-hip claim is a category error.

König, Oesser, Scharla and Zdzieblik randomised 131 postmenopausal women with age-related bone-mineral-density reduction to 5 g/day specific collagen peptides or placebo for twelve months (König et al., 2018). One hundred two completed; the published analysis is intention-to-treat (mean age 64.3 years; spine T-score −2.4; femoral-neck T-score −1.4). Spine T-score change was +0.1 versus −0.03 (ANCOVA P = 0.030). Femoral-neck T-score change was +0.09 versus −0.01 (P = 0.003). P1NP rose in the peptide arm. The absolute T-score deltas are small. Oesser is a coauthor. The trial is not a fracture trial. Those three sentences are the result.

Elam and colleagues randomised 39 osteopenic women to a 5 g calcium–collagen chelate providing 500 mg calcium and 200 IU vitamin D, or to 500 mg calcium plus 200 IU vitamin D, for twelve months (Elam et al., 2015). The harvested abstract is cut before numeric bone-mineral-density deltas. Those percentages are not invented here. The comparison is a blend versus calcium and vitamin D, not peptides alone versus nothing.

Daneault reviewed hydrolyzed collagen and bone metabolism; the underlying file is mostly rodent and in-vitro work (Daneault et al., 2017). Sun, Yang, Teng and Xia meta-analysed collagen-peptide trials, often combined with calcium and vitamin D, and reported increases in femoral-neck and spine bone-mineral density with I² = 80.1% for the density endpoint, bone-turnover standardised mean differences of 0.40–0.58, and a muscle standardised mean difference of 0.60 whose confidence interval begins at 0.05 (Sun et al., 2025). High heterogeneity and frequent co-administration of calcium and vitamin D are the controlling qualifications. Brueckheimer, Costa Silva, Rodrigues and Zague, restricting to type-I hydrolysate, found bone conclusions limited (Brueckheimer et al., 2025).

Adult cartilage collagen residence time is measured in decades; normal disc collagen half-life is on the order of a century (Verzijl et al., 2000; Sivan et al., 2008). Bone remodels faster than that, which is why a twelve-month T-score trial can exist at all. Speed of turnover is not a licence to treat a 0.1 T-score movement as fracture prevention.

20Muscle mass and protein supplementation

This is the section in which the class claim dies. Muscle protein synthesis is gated by essential amino acids, leucine prominent among them (Phillips and Van Loon, 2011; Churchward-Venne et al., 2012; Wolfe et al., 2024). Protein quality is the ability of a food to supply those nine dietary essential amino acids, corrected for digestibility. The current scoring method is the digestible indispensable amino-acid score (FAO Expert Consultation, 2013; Wolfe et al., 2024). Gorissen, Crombag, Senden and Waterval assayed a panel of plant and animal isolates: whey 43% essential amino acids, milk 39%, human muscle 38%, casein 34%, egg 32%, oat, lupin, and wheat about 21–22% (Gorissen et al., 2018). Collagen was not in that panel. A numeric collagen score is not printed here because the 1955 composition paper that would support residue-level percentages has no harvested abstract (Eastoe, 1955). What is verified without that table is the physiology: when collagen peptides are compared with dairy proteins, they lose.

Oikawa, Macinnis, Tripp, McGlory and Phillips ran the load-bearing tracer trial (Oikawa et al., 2020a). Twenty-two older women (about 69 years) consumed 30 g whey protein or 30 g collagen peptides twice daily for six days, with unilateral resistance exercise. Acute myofibrillar protein synthesis on whey was 0.017 ± 0.008%/h at rest and 0.032 ± 0.012%/h after exercise. Collagen reached 0.012 ± 0.013%/h only in the exercised leg. Whey exceeded collagen in both legs (P = 0.02). Integrated synthesis over the feeding period rose with whey and did not rise significantly with collagen. That is a six-day feeding study, not a twelve-week hypertrophy trial, and it is the cleanest independent test of collagen as a muscle protein in the reviewed record.

In a separate crossover in 11 endurance-trained adults, Oikawa compared α-lactalbumin with collagen peptides at 20 g after intensified aerobic work plus 40 g pre-sleep (Oikawa et al., 2020b). Lactalbumin raised plasma leucine and tryptophan relative to collagen (P < 0.001 for both). Myofibrillar synthesis was higher with lactalbumin. The harvested abstract reports a 13% difference before truncation. The trial is small. The direction is not ambiguous. Collagen was used, correctly, as the low-leucine, low-tryptophan comparator.

Jacinto, Nunes, Gorissen and Capel then asked whether free leucine could rescue collagen during ten weeks of supervised resistance training (Jacinto et al., 2022). Twenty-two untrained young adults received 35 g whey (3.0 g leucine) or 35 g collagen peptides (1.0 g leucine plus 2.0 g free leucine) after workouts and on non-training evenings. Vastus-lateralis thickness rose 8.4 ± 2.5% on whey and 5.6 ± 2.6% on leucine-matched collagen. Biceps-brachii thickness rose 10.1 ± 3.8% on whey, with a smaller collagen change. Free leucine did not convert collagen into whey. Churchward-Venne, Burd, Mitchell and West had already shown that 25 g whey kept myofibrillar synthesis elevated three to five hours after exercise, whereas 6.25 g whey plus leucine, or plus essential amino acids without matching leucine, did not (Churchward-Venne et al., 2012). A low-quality protein remains a low-quality protein when a limiting amino acid is sprinkled on top.

Robberechts, Poffé, Ampe and Bogaerts tested the brochure sentence that adding collagen to whey should improve recovery (Robberechts et al., 2024). Twenty-two fit men completed three weeks of unilateral eccentric training on 45 g/day whey or on 25 g whey plus 20 g collagen peptides. Maximal voluntary isometric and dynamic contraction fell about 10% at 48 hours in both arms. Creatine kinase doubled in both arms. Substituting part of the whey did not improve damage indices or functional recovery. Athletes already eating a high-quality protein do not gain a recovery advantage by replacing some of it with collagen.

The positive fat-free-mass literature is a different object and a different author cluster. Zdzieblik, Oesser, Baumstark and Gollhofer randomised 53 sarcopenic men (mean age 72.2 years) to 12 weeks of guided resistance training three times weekly plus 15 g/day collagen peptides or a silica placebo (Zdzieblik et al., 2015). Fat-free mass, fat mass, bone mass, isokinetic quadriceps strength, and a one-leg stabilisation test all moved with training; the abstract states that the effect was more pronounced in the peptide arm. Phillips, Tipton, van Loon, Verdijk and Paddon-Jones answered in the same journal the following year. Their letter is titled for what they thought of the magnitude: “Exceptional body composition changes attributed to collagen peptide supplementation and resistance training in older sarcopenic men” (Phillips et al., 2016). The letter has no abstract body in the reviewed record. The objection stands as a published, independent, protein-metabolism reading of that trial and must be carried beside it.

Jendricke, Centner, Zdzieblik and Gollhofer randomised 77 premenopausal women to 12 weeks of resistance training three days per week plus 15 g/day specific collagen peptides or placebo (Jendricke et al., 2019). Bioimpedance fat-free-mass percentage and hand-grip rose more in the peptide arm. Oertzen-Hagemann, Kirmse, Eggers and Pfeiffer assigned 25 young men to 15 g/day specific collagen peptides or a non-caloric placebo after hypertrophy sessions, three times weekly for twelve weeks (Oertzen-Hagemann et al., 2019). Body mass and fat-free mass increased versus placebo; strength rose in both groups. Kirmse, Oertzen-Hagemann, de Marées and Bloch randomised 57 recreationally active men to 15 g/day collagen peptides or placebo across the same twelve-week, three-sessions-per-week design (Kirmse et al., 2019). Fat-free mass rose versus placebo. Type II fibre cross-sectional area rose in both groups without a between-group difference. The authors themselves noted that the fat-free-mass increment was not reflected in fibre area. Connective tissue, glycogen, and water all sit inside a composition machine. A glycine-rich hydrolysate can move that machine without building myosin.

Phillips and Van Loon’s consensus range for athletes is 1.3–1.8 g·kg⁻¹·d⁻¹ as three to four meals, higher in heavy blocks, with leucine in a position of prominence (Phillips and Van Loon, 2011). The NIH Office of Dietary Supplements exercise-and-performance fact sheet cites 1.2–2.0 g/kg for athletes and does not list collagen among ingredients with performance evidence. Bischof’s training meta-analysis is mixed (Bischof et al., 2024). Sun’s muscle standardised mean difference of 0.60 begins its confidence interval at 0.05 and inherits the orbit file (Sun et al., 2025).

The class claim that collagen peptides are a complete muscle-building protein, or a substitute for adequate essential-amino-acid intake, is contradicted by the best independent tracer and hypertrophy work in the reviewed record. That verdict is strongly supported as a negative. Collagen peptides can still contribute glycine, proline, and hydroxyproline to a diet that already meets essential-amino-acid needs. That is a different claim, and it is not a quality claim.

Outcome claims attach to the preparation that was swallowed, not to the English word “collagen.” Most of the positive skin, joint, tendon-morphology, bone-density, and fat-free-mass trials in the reviewed record either name a trademarked mix or come from one manufacturer-adjacent author cluster. The independent protein-quality trials, which test collagen as a protein, go the other way.

21Branded versus generic

Four materials share a store aisle and almost no analytical identity. Native fibrillar collagen is the triple-helical tissue protein whose structure is established (Shoulders and Raines, 2009; Ricard-Blum, 2011). Culinary or pharmaceutical gelatin is heat-denatured collagen, still high-molecular-weight; it is what Shaw and Lis administered with vitamin C before loading (Shaw et al., 2017; Lis and Baar, 2019). Generic collagen hydrolysate is enzymatically cut gelatin with a variable peptide map; it is what Iwai used for absorption and what Oikawa, Jacinto, and Robberechts used as a protein comparator (Iwai et al., 2005; Oikawa et al., 2020a; Jacinto et al., 2022; Robberechts et al., 2024). Branded bioactive collagen peptides are patented, composition-defined hydrolysate fractions. Undenatured type II collagen is a fifth object if one is counting commercial SKUs, and it is not a peptide mix at all (Martínez-Puig et al., 2023; Van Vijven et al., 2012). Marketing collapses the set. The trials do not.

Van Vijven already split undenatured collagen, gelatin, and hydrolysate in 2012. Martínez-Puig restated the chemical consequence in 2023: native collagen is hypothesised to act by epitope recognition; hydrolysates, if they act, act as peptides. A 40 mg UC-II capsule and a 10 g type-I hydrolysate drink are not two doses of one drug. They are two products.

Amounts are reported study parameters. They are not use instructions. A store-brand powder without a published peptide map is not any row in this table.

What generic collagen has been tested for is protein quality, not beauty or joints. Oikawa’s older women lost to whey for myofibrillar synthesis on 30 g twice daily for six days (Oikawa et al., 2020a). Oikawa’s endurance-trained adults lost to α-lactalbumin for myofibrillar synthesis during intensified cycling (Oikawa et al., 2020b). Jacinto’s untrained adults lost to whey for muscle thickness over ten weeks even when free leucine was added to the collagen arm (Jacinto et al., 2022). Robberechts’s fit men gained no recovery advantage when 20 g of a 45 g whey allotment was replaced with collagen peptides during eccentric training (Robberechts et al., 2024). Kirmse’s young men gained no further connective-tissue or myofibrillar fractional synthesis from 15 g hydrolyzed collagen twice daily during a week of heavy training (Kirmse et al., 2024). That is the generic-product evidence that exists. It is not a skin programme and it is not an osteoarthritis programme.

Batch non-identity is strongly supported at the bench. Schadow and colleagues showed that different collagen-hydrolysate batches differ in peptide composition and in how they move human osteoarthritic cartilage and synoviocytes; some Peptan-class materials appear in that set (Schadow et al., 2013; Schadow et al., 2017; Simons et al., 2018). Inoue, Sugihara and Wang showed the same fact on a face: two hydrolysates with different Pro-Hyp and Hyp-Gly contents were not equal on moisture, elasticity, wrinkles, or roughness (Inoue et al., 2016). A reader who says “all hydrolysates yield Pro-Hyp, so they should act alike” has confused a shared circulating dipeptide (Iwai et al., 2005) with an identical oligopeptide dose, molecular-weight distribution, and co-ingredient list. Appearance is common. Composition is not.

Species is a catalogue variable, not a demonstrated active variable. Evans used a freshwater marine hydrolysate (Evans et al., 2021). Kim used a fish LMWCP (Kim et al., 2018). Schauss and the UC-II papers used chicken sternum (Schauss et al., 2012; Crowley et al., 2009; Lugo et al., 2016). Iwai used porcine and chicken gelatin hydrolysates (Iwai et al., 2005). Once peptide map, amount, duration, and co-ingredients differ, species is not isolable. Marine versus bovine is a marketing contrast until someone holds those other variables still.

22Industry-funding ledger

Sponsorship is a property of a paper, not a reason to discard a result. It is a reason not to treat twenty orbit papers as twenty independent discoveries of collagen. Abstracts under-detect the property. A keyword scan of the reviewed recorded abstract set found twenty brand or funding-string hits and included false positives (topical testosterone Fortigel; yoga BodyBalance). Most Gelita-orbit papers do not say “Gelita” in the abstract. Full conflict-of-interest statements from publisher XML were not retrieved for this review. A sentence of the form “X percent of trials were paid by industry” would be an invention. It is not printed.

What can be seen is labelled. Named in abstract means a brand or manufacturer string is in the PubMed abstract. Orbit means the recurring Oesser, Schunck, König, Zdzieblik, Jendricke, Jerger, Dressler set, or the phrase “specific (bioactive) collagen peptides,” with the brand word absent from the abstract. Their conflicts are unverified, not proven grants. Manufacturer-class means a trademarked hydrolysate or extract (UC-II, BioCell, CH-Alpha, ELASTEN, Vinh Wellness, LMWCP). Independent means an academic protein-metabolism or imaging group and no product name in the abstract. Reviews inherit the file they pool.

The orbit cluster publishes the fat-free-mass, T-score, knee visual-analogue, ankle, and tendon-area papers, usually without the brand word in the abstract: Proksch and Schunck 2014 on elasticity; Zdzieblik, Oesser and König 2015 on sarcopenic fat-free mass; Zdzieblik, Oesser and König 2017 on activity-related knee pain; König, Oesser and Zdzieblik 2018 on bone T-scores; Dressler, Zdzieblik and Oesser 2018 on ankle scores; Jendricke, Zdzieblik and König 2019 on women’s fat-free mass; Oertzen-Hagemann and Kirmse 2019 on the muscle proteome; Kirmse and Oertzen-Hagemann 2019 on fat-free mass versus fibre area; Jerger 2022 and 2023 on Achilles and patellar cross-sectional area; and Kirmse 2024 on connective-tissue fractional synthesis, which is a null (Proksch et al., 2014a; Zdzieblik et al., 2015; Zdzieblik et al., 2017; König et al., 2018; Dressler et al., 2018; Jendricke et al., 2019; Oertzen-Hagemann et al., 2019; Kirmse et al., 2019; Jerger et al., 2022; Jerger et al., 2023; Kirmse et al., 2024). Treating each of those papers as an independent replication of “collagen” is the category error this part exists to prevent. The inference that the cluster is one industrial peptide platform, published repeatedly in Nutrients and sports-nutrition journals, is labelled as inference. Upgrading any orbit row to a confirmed manufacturer grant would require publisher conflict statements that this review did not retrieve.

The independent counterweight is small and consistent. McAlindon’s dGEMRIC pilot and Bruyère’s 1.2 g responder trial used a manufacturer hydrolysate but are not the Gelita author set (McAlindon et al., 2011; Bruyère et al., 2012). Shaw and Lis tested gelatin or hydrolysate plus vitamin C plus jumping as a biomarker protocol (Shaw et al., 2017; Lis and Baar, 2019). Oikawa, Jacinto, and Robberechts tested collagen as a protein and found it inferior or non-additive (Oikawa et al., 2020a; Oikawa et al., 2020b; Jacinto et al., 2022; Robberechts et al., 2024). Phillips, Tipton, van Loon, Verdijk and Paddon-Jones rejected the magnitude of Zdzieblik 2015 in a letter (Phillips et al., 2016). That is the file that does not treat collagen as a hero product.

Reviews inherit whatever they include. Choi, de Miranda, and Pu pool branded skin mixes (Choi et al., 2019; de Miranda et al., 2021; Pu et al., 2023). Pu notes bias. García-Coronado, Van Vijven, and Honvo disagree on osteoarthritis instruments (García-Coronado et al., 2019; Van Vijven et al., 2012; Honvo et al., 2020). Moskowitz reports the null multinational hydrolysate trial (Moskowitz, 2000). Bello and Oesser’s osteoarthritis narrative is written with Oesser, a hydrolysate-industry scientist, as coauthor (Bello and Oesser, 2006). Khatri and Bischof already describe the training literature as mixed (Khatri et al., 2021; Bischof et al., 2024). Sun’s bone-and-muscle meta-analysis carries I² = 80.1% on bone-mineral density (Sun et al., 2025). Martínez-Puig is useful on composition and may be industry-adjacent as a venue; that conflict is unverified (Martínez-Puig et al., 2023).

23What would generalize

Generalisation is a design, not a hope. A result on VERISOL 2.5 g/day for eight weeks in 114 women aged 45–65 years generalises to that peptide map, that amount, that duration, and that population (Proksch et al., 2014b). It does not generalise to an unassayed bovine powder in a man of 30. A result on TENDOFORTE plus calf loading in 20 people with mid-portion Achilles tendinopathy generalises, provisionally, to that named product plus that loading programme (Praet et al., 2019). It does not generalise to gelatin gummies, and Lis already showed that a gummy did not raise the collagen-synthesis marker that gelatin and hydrolysate tended to raise (Lis and Baar, 2019). A result on 5 g/day specific collagen peptides plus heavy resistance training in healthy men generalises to tendon cross-sectional area in that platform (Jerger et al., 2022; Jerger et al., 2023). It does not generalise to patient-reported tendinopathy care, and it does not overthrow Heinemeier’s bomb-¹⁴C stillness of the healthy Achilles core (Heinemeier et al., 2013).

What would count as a class effect is easy to state and absent from the reviewed record. A pre-registered, independently funded randomised trial of an off-the-shelf hydrolysate, composition-assayed before unblinding, against both an isonitrogenous whey or essential-amino-acid control and the branded mix that produced the original signal, on the same endpoint, in a population that matches the claim. That trial is not in the reviewed record. Until it is, the burden of proof sits on anyone who asserts a class-wide skin, joint, tendon, bone, or muscle benefit.

Two weaker substitutes are often offered. The first is meta-analysis. Pu’s 26-trial hydration and elasticity estimates are real pooled statistics on the literature that exists (Pu et al., 2023). They remain composition-blind. Pooling VERISOL, LMWCP, a marine hydrolysate, and ELASTEN does not create a test of grocery collagen; it conceals the fact that grocery collagen was never tested. The second substitute is shared pharmacokinetics. Pro-Hyp appears after many oral gelatin hydrolysates (Iwai et al., 2005; Yazaki et al., 2017). Shared appearance of one dipeptide is compatible with Inoue’s demonstration that two hydrolysates are not equal on the face, and with Schadow’s demonstration that two batches are not equal on osteoarthritic cartilage (Inoue et al., 2016; Schadow et al., 2013). The warrant is simple: the swallowed composition is the intervention.

Industry funding of the only trials large enough to run is a structural fact, not an excuse. If manufacturers are the only parties who will randomise a named mix against placebo on a cutometer or a WOMAC, the article must say so, and must overweight the few independent head-to-heads. Those head-to-heads are the protein-quality file. They are negative for the muscle-building class claim (Oikawa et al., 2020a; Jacinto et al., 2022). They do not retire a VERISOL wrinkle-volume difference. They do prevent anyone from sliding from that difference to “collagen builds muscle” or “any collagen powder is an anti-wrinkle treatment.”

The standing inference of this part is the thesis restated at industrial scale. Native fibrillar collagen, culinary gelatin, generic hydrolysate, and branded bioactive peptide compositions are four materials. UC-II is a fifth commercial object and a different hypothesis. Human trials that move instrumental skin, joint, or body-composition endpoints almost all administer a named mix. Those results do not license a claim that any collagen powder is a complete muscle-building protein, a clinically meaningful anti-wrinkle treatment, or a substitute for adequate essential-amino-acid intake.

Part Five therefore does not reopen the industrial file. It asks the six questions the aisle already answers in advertising language, and it keeps the safety and sibling-scope sentences from being mistaken for an efficacy upgrade. The adversarial table at the end of the Apparatus is the same verdict written once, not a seventh claim.

The six answers are already in the body. Part Five writes them as questions so a reader can refuse a brochure sentence without rereading Parts Three and Four. Safety is a short file because this review was not a pharmacovigilance programme. What would change the verdict is a design, not a hope, and it is the same design Part Four already named: composition-assayed, independently funded, quality-matched, and repeated.

Collagen biology is not in doubt. The commercial object sold as collagen peptides is. Six questions decide whether the aisle’s three class claims survive the file. They do not. Safety, such as it is reported, is the safety of a food protein, not the safety of a disease-modifying drug.

24 Adversarial review

Q1. Is collagen superior to adequate protein intake?

No. For myofibrillar protein synthesis and hypertrophy it is inferior to adequate essential-amino-acid intake from whey or mixed high-quality protein. It is not a complete protein in the FAO sense. The grade is strongly supported as a negative for the superiority claim.

Muscle protein synthesis is gated by essential amino acids, leucine prominent among them (Phillips and Van Loon, 2011; Churchward-Venne et al., 2012; Wolfe et al., 2024). Quality is a digestible indispensable amino-acid score, not grams of nitrogen (FAO Expert Consultation, 2013; Wolfe et al., 2024). Head-to-head, whey exceeds collagen for acute and six-day myofibrillar synthesis in older women taking 30 g twice daily (Oikawa et al., 2020a). α-Lactalbumin exceeds collagen for myofibrillar synthesis during intensified aerobic work at 20 g post-exercise plus 40 g pre-sleep (Oikawa et al., 2020b). Whey exceeds leucine-matched collagen for ten-week muscle thickness in untrained adults taking 35 g (Jacinto et al., 2022). Adding 20 g collagen peptides in place of part of a 45 g whey allotment does not improve eccentric recovery (Robberechts et al., 2024). Fifteen grams of hydrolyzed collagen twice daily does not raise muscle connective-protein synthesis during one week of heavy resistance training (Kirmse et al., 2024). If “superior to adequate protein” means better muscle remodelling than a diet that already meets essential-amino-acid needs, those trials falsify it.

The objection is Zdzieblik 2015 and the later specific-peptide plus resistance-training papers that report extra fat-free mass versus silica or non-caloric placebo (Zdzieblik et al., 2015; Jendricke et al., 2019; Kirmse et al., 2019). The reply is already in print. Phillips, Tipton, van Loon, Verdijk and Paddon-Jones called the 2015 body-composition change exceptional (Phillips et al., 2016). Kirmse 2019 raised fat-free mass without a between-group difference in type II fibre area (Kirmse et al., 2019). Fat-free mass is not myosin. Those trials also lack a whey arm. The hidden premise to reject is that protein is protein. Collagen can add glycine, proline, and hydroxyproline to an already complete diet. That is a different claim.

Q2. Are skin benefits clinically meaningful?

They are statistically detectable on devices for named preparations. They have not been shown to be clinically meaningful in the sense of a patient-important, independently replicated dermatologic outcome. The instrumental signal is emerging. The class-wide anti-wrinkle treatment claim is speculative.

VERISOL 2.5 g/day for eight weeks reduced eye-wrinkle volume versus placebo, quoted as 20% at eight weeks, in 114 women aged 45–65 years (Proksch et al., 2014b). Elasticity rose in the 69-woman specific-hydrolysate trial; moisture did not, overall (Proksch et al., 2014a). Asserin reported hydration and dermal density (Asserin et al., 2015). Inoue showed that a higher-dipeptide hydrolysate outperformed a lower-dipeptide hydrolysate (Inoue et al., 2016). Kim’s 1,000 mg/day LMWCP moved hydration and some wrinkle and elasticity parameters in 64 participants over twelve weeks (Kim et al., 2018). Evans’s marine hydrolysate reduced a wrinkle score 35% from baseline (P = 0.035) in women aged 45–60 years, with elasticity confined to a younger subgroup (Evans et al., 2021). Bolke’s ELASTEN win is a blend of 2.5 g peptides plus vitamin C, zinc, biotin, and vitamin E and cannot be attributed to collagen (Bolke et al., 2019). Choi, de Miranda, and Pu then pool these products (Choi et al., 2019; de Miranda et al., 2021; Pu et al., 2023). Pu notes bias and source-and-duration dependence. Endpoints are corneometry, cutometry, PRIMOS, VISIA, and ultrasound — not a drug-style clinician global or a five-year photoageing hard outcome.

The objection is that consumers care about hydration and wrinkle scores, and that Pu assembled 26 randomised trials in 1,721 subjects. Large and pooled is not independent and generalisable. The file is women, short, sponsor-heavy, and composition-blind. Device-significant millimetre changes on a patented mix do not license a class-wide anti-wrinkle treatment for a powder the trial never swallowed.

Q3. How much evidence is industry sponsored?

Most of the positive outcome file is manufacturer-named or manufacturer-adjacent. The independent protein-quality file is not. A numeric “X percent of trials were paid by industry” is unverified, because conflict statements were not retrieved from publisher bodies, and is not printed. The grade for that detection limit is strongly supported.

Abstracts name VERISOL, TENDOFORTE, ELASTEN, CH-Alpha, BioCell, UC-II, Vinh Wellness Collagen, and LMWCP. A second cluster, writing “specific collagen peptides” with Oesser, König, Zdzieblik, or Jerger as markers, produces the fat-free-mass, T-score, knee visual-analogue, and tendon-area papers without the brand word in the abstract. Their conflicts are unverified, not proven grants. A keyword scan of 370 abstracts found only twenty brand or funding hits and included false positives. Abstracts under-detect sponsorship. The independent negatives cluster in Phillips, Oikawa, Jacinto, Robberechts, and Baar.

The objection is that industry funds the only trials large enough to run. Then the article must say so, and must overweight the few independent head-to-heads. Sponsorship is not a reason to discard a VERISOL wrinkle-volume difference. It is a reason not to count twenty orbit papers as twenty independent discoveries of collagen.

Q4. Do branded preparations generalize to all collagen?

No. The grade is strongly supported.

Native type II at 40 mg (UC-II) is a different molecule and a different hypothesised mechanism than grams of type-I hydrolysate (Martínez-Puig et al., 2023; Lugo et al., 2016). Hydrolysate batches differ in peptide maps and in osteoarthritic-cartilage bioactivity (Schadow et al., 2013; Simons et al., 2018). Two hydrolysates with different Pro-Hyp and Hyp-Gly contents are not equal on skin instruments (Inoue et al., 2016). Positive randomised trials name VERISOL, TENDOFORTE, BioCell, CH-Alpha, ELASTEN, or “specific collagen peptides.” Generic collagen, when used as a protein control, loses (Oikawa et al., 2020a; Jacinto et al., 2022). A result attaches to the swallowed composition. Grocery collagen without that map is an untested product.

The objection is that all hydrolysates yield Pro-Hyp, so they should act alike (Iwai et al., 2005). Appearance of a dipeptide is common. Dose of specific oligopeptides, molecular-weight distribution, and co-ingredients are not. Inoue already showed composition dependence on the same instruments the aisle quotes.

Q5. Are exercise plus tendon findings replicated?

Not as clinical rehabilitation. Morphology in healthy tendons has a same-laboratory echo. The tendinopathy pilot has not been independently repeated. The grade for clinical replication is unmeasured. The grade for a preliminary, platform-specific morphological signal is emerging.

Shaw 2017 is eight men, gelatin plus vitamin C plus skipping, a collagen-synthesis marker and an engineered ligament (Shaw et al., 2017). Lis 2019 is ten men; the same marker tended to rise about 20% with gelatin or hydrolysate and not with a gummy (Lis and Baar, 2019). Praet 2019 is twenty patients with mid-portion Achilles tendinopathy, TENDOFORTE plus loading, a first-period VISA-A advantage of 12.6 versus 5.3 points (Praet et al., 2019). Jerger 2022 and 2023 are healthy men, 5 g/day specific collagen peptides plus heavy resistance training, greater Achilles or regional patellar cross-sectional area, with stiffness and strength not different from placebo (Jerger et al., 2022; Jerger et al., 2023). Miyamoto 2025 is sedentary men, 10 g/day collagen peptides, stiffness and rate of torque development up, cross-sectional area unchanged — a different endpoint (Miyamoto et al., 2025). Kirmse 2024 is no added connective fractional synthesis from hydrolysate during training (Kirmse et al., 2024). Heinemeier 2013 is a healthy Achilles core that is not renewed after growth (Heinemeier et al., 2013). Heinemeier 2018 is tendinopathic matrix with years of abnormal turnover before pain (Heinemeier et al., 2018). Hijikata’s Achilles hydrolysate trial was not recovered and is not cited.

Replication means a second independent group, patients not students, the same clinical endpoint. Jerger is same-platform morphology, not a Praet replication. Shaw and Lis are biomarkers in samples smaller than twelve. Loading changes tendon, and a specific peptide may enlarge cross-sectional area in healthy men. That is not replicated tendinopathy care, and it does not overthrow bomb-¹⁴C stillness of the healthy core.

Q6. Are muscle-building claims overstated?

Yes. The grade is strongly supported.

Marketing treats collagen as a muscle protein. Collagen is essential-amino-acid-poor relative to the proteins that raise myofibrillar synthesis. Gorissen’s panel did not include collagen; the physiological contrast is Oikawa’s leucine and tryptophan rise on lactalbumin versus collagen, and Oikawa’s whey-versus-collagen tracer in older women (Gorissen et al., 2018; Oikawa et al., 2020a; Oikawa et al., 2020b; Wolfe et al., 2024). Direct myofibrillar-synthesis and hypertrophy head-to-heads are negative for collagen versus dairy (Oikawa et al., 2020a; Oikawa et al., 2020b; Jacinto et al., 2022; Robberechts et al., 2024). The positive fat-free-mass literature is orbit, DXA or bioimpedance, and, in the one fibre study, dissociated from type II area (Kirmse et al., 2019). The flagship sarcopenia trial was publicly challenged (Phillips et al., 2016). Khatri and Bischof already say the training literature is mixed (Khatri et al., 2021; Bischof et al., 2024). A muscle-building claim requires myofibrillar synthesis or fibre hypertrophy against a quality-matched control. Collagen fails that test. Older sarcopenic men gained fat-free mass and strength with 15 g peptides plus resistance training versus silica (Zdzieblik et al., 2015). Versus silica is not versus whey. Magnitude contested is not a licence for the class.

25Safety

This harvest is an efficacy and identity file, not a pharmacovigilance programme. The safety statements that can be made are the statements the trials actually recorded. They are consistent with a food protein. They are not a long-term disease-modifying-drug dossier. No amount, route, or schedule is recommended.

Moskowitz, reviewing pharmaceutical-grade collagen hydrolysate, wrote that hydrolyzed gelatin products have long been used in pharmaceuticals and foods and are generally recognized as safe food products by regulatory agencies, and that clinical use was associated with minimal adverse effects, mainly gastrointestinal fullness or unpleasant taste (Moskowitz, 2000). Bruyère’s six-month trial of 1,200 mg/day hydrolysate in 200 people aged 50 years or older found no significant difference in security and tolerability versus placebo (Bruyère et al., 2012). Kim recorded no adverse symptoms attributed to 1,000 mg/day LMWCP over twelve weeks in 64 participants (Kim et al., 2018). Praet reported no adverse events in the twenty-person TENDOFORTE crossover (Praet et al., 2019). Schauss reported a tolerability profile comparable to placebo for 2 g/day BioCell over 70 days in 80 people with hip or knee osteoarthritis (Schauss et al., 2012). Inoue evaluated safety by blood test in the high-versus-low dipeptide comparison (Inoue et al., 2016). Benito-Ruiz’s six-month, 10 g/day, 250-person osteoarthritis trial is framed as a food-ingredient study and concludes that the hydrolysate “is safe and effective” in the authors’ words (Benito-Ruiz et al., 2009). Those are trial-level tolerability notes. They are not a pooled adverse-event rate, and a pooled rate is not invented here.

What the reviewed record does not contain belongs beside those sentences. There is no dedicated long-term safety randomised programme for branded bioactive peptides, generic hydrolysate, gelatin, or UC-II. There is no fracture-endpoint trial whose safety follow-up could double as an efficacy follow-up (König et al., 2018). There is no harvest paper that isolates allergenicity by source species. Bovine, porcine, marine, and chicken-sternal products appear across the trial matrix (Iwai et al., 2005; Kim et al., 2018; Evans et al., 2021; Schauss et al., 2012; Crowley et al., 2009). A person with a documented allergy to a source tissue is outside the evidential claim of a brochure that says “collagen is just protein.” That is identity, not a scare sentence.

UC-II remains a separate safety object because it is a separate molecule. Crowley, Lugo 2013, and Lugo 2016 report tolerability language for 40 mg/day undenatured type II over 90 to 180 days (Crowley et al., 2009; Lugo et al., 2013; Lugo et al., 2016). Those papers do not make a hydrolysate safe, and a hydrolysate trial does not make UC-II safe. ELASTEN’s twelve-week blend trial adds vitamin C, zinc, biotin, and vitamin E to 2.5 g peptides (Bolke et al., 2019). Blend tolerability is blend tolerability.

Culture concentrations of Pro-Hyp at 200 nmol/mL are not human safety data (Shigemura et al., 2009; Ohara et al., 2010). Mouse cartilage accumulation of labelled gelatin hydrolysate is not human safety data (Oesser et al., 1999). The NIH Office of Dietary Supplements exercise-and-performance fact sheet does not list collagen among ingredients with performance evidence; it is silent as a safety article. Absence of a listed ergogenic claim is not a toxicity finding.

26What would change the verdict

The thesis stands until the file changes. Three kinds of paper would move it. None of them is in the reviewed record.

The first is a generalisation trial. A pre-registered, independently funded randomised comparison of an off-the-shelf hydrolysate, composition-assayed before unblinding, against the branded mix that produced the original signal and against an isonitrogenous whey or essential-amino-acid control, on the same endpoint, in a population that matches the claim. If a grocery powder matches VERISOL on eye-wrinkle volume, or matches a specific collagen peptide on Achilles cross-sectional area, the branded-versus-generic wall moves. If it matches whey on myofibrillar synthesis, the protein-quality verdict moves. The present independent protein trials run the other way (Oikawa et al., 2020a; Jacinto et al., 2022).

The second is an endpoint upgrade. A fracture-endpoint randomised trial would retire the unmeasured grade on bone (König et al., 2018; Sun et al., 2025). An independently replicated VISA-A trial in Achilles tendinopathy, by a group that is not the Praet–TENDOFORTE pairing and not the Jerger morphology laboratory, would retire the unreplicated-care grade (Praet et al., 2019). A dermatologist-global or multi-year photoageing primary, on a composition-defined product, would retire the device-only grade on skin (Proksch et al., 2014b; Pu et al., 2023). A disease-modifying osteoarthritis outcome would retire Honvo’s mapping result (Honvo et al., 2020). A published digestible indispensable amino-acid score for a defined collagen hydrolysate, read from a composition table rather than inferred from Oikawa’s leucine and tryptophan contrast, would replace the present refusal to print a numeric score (Eastoe, 1955; Wolfe et al., 2024). Residue-level tryptophan grams are not printed until that table is read.

The third is a conflict file. Publisher or ClinicalTrials.gov sponsor fields that either clear or convict the orbit set would retire the unverified-conflict label on the Oesser–König–Zdzieblik–Jerger cluster. Keyword hits in abstracts are not that file. Twenty hits including false Fortigel and BodyBalance strings are a detection limit, not a percentage of industry dollars.

Lesser papers would not move the verdict. Another orbit fat-free-mass trial versus silica would not, because Phillips already objected to the magnitude and Kirmse already dissociated fat-free mass from type II fibre area (Phillips et al., 2016; Kirmse et al., 2019). Another meta-analysis that pools named mixes and blends would not, because Choi, de Miranda, and Pu have already pooled them (Choi et al., 2019; de Miranda et al., 2021; Pu et al., 2023). Another Pro-Hyp pharmacokinetic time course would not, because appearance was established in 2005 and is not an outcome (Iwai et al., 2005). Another UC-II symptom trial would not change a hydrolysate verdict, because UC-II is not a collagen-peptide result (Lugo et al., 2016). Another ELASTEN-style blend would not attribute an effect to peptides (Bolke et al., 2019).

Until those papers exist, the controlling sentences remain the ones this article was written to keep apart. Collagen biology is established. The supplement class is not a complete protein and is not superior to adequate essential-amino-acid intake for myofibrillar synthesis. Skin and some joint or tendon signals exist for specific branded preparations at instrumental or symptom scales. They do not generalise to all collagen powders. A large fraction of the outcome literature is manufacturer-named or manufacturer-adjacent. Exercise-plus-tendon findings are not independently replicated as clinical rehabilitation. Muscle-building claims for the class are overstated. The scoop is several products. The helix is not one of them.

27Research matrices

The tables restate numbers already argued. They do not add effect sizes. Amounts are published trial parameters, not recommendations. Empty cells are absences, not zeroes.

Table A. Outcome-trial ledger

Table A. Human outcome rows from the reviewed record. A positive skin or joint cell is not a class effect. Check the material column before any generalisation. Full 47-row ledger is in the project evidence packet; this table is the reader-facing extract.

Table B. Skin, joint, tendon, bone

Table B. Domain summary. UC-II rows are excluded from the peptide grade.

Table C. Branded versus generic

Table C. Outcome claims attach to the preparation that was swallowed.

Table D. Protein-quality comparison

Table D. Protein quality is not connective-tissue bioactivity.

Table E. Industry-funding ledger

Table E. No invented percentage of industry dollars. Abstracts under-detect sponsorship. Publisher conflict bodies were not retrieved in this pass. Table F, the Adversarial resolution, is printed after section 30 so that the six answers sit with the series footer rather than being reread as a seventh claim.

28References

1. Al-Atif H. Collagen Supplements for Aging and Wrinkles: A Paradigm Shift in the Fields of Dermatology and Cosmetics. Dermatol Pract Concept. 2022;12(1):e2022018.
   PMID 35223163 · doi:10.5826/dpc.1201a18 · PMC8824545
2. Asserin J, Lati E, Shioya T, Prawitt J. The effect of oral collagen peptide supplementation on skin moisture and the dermal collagen network: evidence from an ex vivo model and randomized, placebo-controlled clinical trials. J Cosmet Dermatol. 2015;14(4):291-301.
   PMID 26362110 · doi:10.1111/jocd.12174
3. Balshaw TG, Funnell MP, McDermott E, Maden-Wilkinson TM, Abela S, Quteishat B, et al.. The effect of specific bioactive collagen peptides on function and muscle remodeling during human resistance training. Acta Physiol (Oxf). 2023;237(2):e13903.
   PMID 36433662 · doi:10.1111/apha.13903 · PMC10078466
4. Bella J, Brodsky B, Berman HM. Hydration structure of a collagen peptide. Structure. 1995;3(9):893-906.
   PMID 8535783 · doi:10.1016/S0969-2126(01)00224-6
5. Bella J. Collagen structure: new tricks from a very old dog. Biochem J. 2016;473(8):1001-25.
   PMID 27060106 · doi:10.1042/BJ20151169
6. Bello AE, Oesser S. Collagen hydrolysate for the treatment of osteoarthritis and other joint disorders: a review of the literature. Curr Med Res Opin. 2006;22(11):2221-32.
   PMID 17076983 · doi:10.1185/030079906X148373
7. Benito-Ruiz P, Camacho-Zambrano MM, Carrillo-Arcentales JN, Mestanza-Peralta MA, Vallejo-Flores CA, Vargas-López SV, et al.. A randomized controlled trial on the efficacy and safety of a food ingredient, collagen hydrolysate, for improving joint comfort. Int J Food Sci Nutr. 2009;60 Suppl 2:99-113.
   PMID 19212858 · doi:10.1080/09637480802498820
8. Bischof K, Moitzi AM, Stafilidis S, König D. Impact of Collagen Peptide Supplementation in Combination with Long-Term Physical Training on Strength, Musculotendinous Remodeling, Functional Recovery, and Body Composition in Healthy Adults: A Systematic Review with Meta-analysis. Sports Med. 2024;54(11):2865-2888.
   PMID 39060741 · doi:10.1007/s40279-024-02079-0 · PMC11561013
9. Bolke L, Schlippe G, Gerß J, Voss W. A Collagen Supplement Improves Skin Hydration, Elasticity, Roughness, and Density: Results of a Randomized, Placebo-Controlled, Blind Study. Nutrients. 2019;11(10).
   PMID 31627309 · doi:10.3390/nu11102494 · PMC6835901
10. Brueckheimer PJ, Costa Silva T, Rodrigues L, Zague V, Isaia Filho C. The Effects of Type I Collagen Hydrolysate Supplementation on Bones, Muscles, and Joints: A Systematic Review. Orthop Rev (Pavia). 2025;17:129086.
    PMID 39980497 · doi:10.52965/001c.129086 · PMC11842160
11. Bruyère O, Zegels B, Leonori L, Rabenda V, Janssen A, Bourges C, et al.. Effect of collagen hydrolysate in articular pain: a 6-month randomized, double-blind, placebo controlled study. Complement Ther Med. 2012;20(3):124-30.
    PMID 22500661 · doi:10.1016/j.ctim.2011.12.007
12. Campos LD, Santos Junior VA, Pimentel JD, Carregã GLF, Cazarin CBB. Collagen supplementation in skin and orthopedic diseases: A review of the literature. Heliyon. 2023;9(4):e14961.
    PMID 37064452 · doi:10.1016/j.heliyon.2023.e14961 · PMC10102402
13. Chai HJ, Li JH, Huang HN, Li TL, Chan YL, Shiau CY, et al.. Effects of sizes and conformations of fish-scale collagen peptides on facial skin qualities and transdermal penetration efficiency. J Biomed Biotechnol. 2010;2010:757301.
    PMID 20625414 · doi:10.1155/2010/757301 · PMC2896882
14. Choi FD, Sung CT, Juhasz ML, Mesinkovsk NA. Oral Collagen Supplementation: A Systematic Review of Dermatological Applications. J Drugs Dermatol. 2019;18(1):9-16.
    PMID 30681787
15. Churchward-Venne TA, Burd NA, Mitchell CJ, West DW, Philp A, Marcotte GR, et al.. Supplementation of a suboptimal protein dose with leucine or essential amino acids: effects on myofibrillar protein synthesis at rest and following resistance exercise in men. J Physiol. 2012;590(11):2751-65.
    PMID 22451437 · doi:10.1113/jphysiol.2012.228833 · PMC3424729
16. Clark KL, Sebastianelli W, Flechsenhar KR, Aukermann DF, Meza F, Millard RL, et al.. 24-Week study on the use of collagen hydrolysate as a dietary supplement in athletes with activity-related joint pain. Curr Med Res Opin. 2008;24(5):1485-96.
    PMID 18416885 · doi:10.1185/030079908x291967
17. Crowley DC, Lau FC, Sharma P, Evans M, Guthrie N, Bagchi M, et al.. Safety and efficacy of undenatured type II collagen in the treatment of osteoarthritis of the knee: a clinical trial. Int J Med Sci. 2009;6(6):312-21.
    PMID 19847319 · doi:10.7150/ijms.6.312 · PMC2764342
18. Daneault A, Prawitt J, Fabien Soulé V, Coxam V, Wittrant Y. Biological effect of hydrolyzed collagen on bone metabolism. Crit Rev Food Sci Nutr. 2017;57(9):1922-1937.
    PMID 25976422 · doi:10.1080/10408398.2015.1038377
19. de Miranda RB, Weimer P, Rossi RC. Effects of hydrolyzed collagen supplementation on skin aging: a systematic review and meta-analysis. Int J Dermatol. 2021;60(12):1449-1461.
    PMID 33742704 · doi:10.1111/ijd.15518
20. Dressler P, Gehring D, Zdzieblik D, Oesser S, Gollhofer A, König D. Improvement of Functional Ankle Properties Following Supplementation with Specific Collagen Peptides in Athletes with Chronic Ankle Instability. J Sports Sci Med. 2018;17(2):298-304.
    PMID 29769831 · PMC5950747
21. EASTOE JE. The amino acid composition of mammalian collagen and gelatin. Biochem J. 1955;61(4):589-600.
    PMID 13276342 · doi:10.1042/bj0610589 · PMC1215839
22. Elam ML, Johnson SA, Hooshmand S, Feresin RG, Payton ME, Gu J, et al.. A calcium-collagen chelate dietary supplement attenuates bone loss in postmenopausal women with osteopenia: a randomized controlled trial. J Med Food. 2015;18(3):324-31.
    PMID 25314004 · doi:10.1089/jmf.2014.0100
23. Evans M, Lewis ED, Zakaria N, Pelipyagina T, Guthrie N. A randomized, triple-blind, placebo-controlled, parallel study to evaluate the efficacy of a freshwater marine collagen on skin wrinkles and elasticity. J Cosmet Dermatol. 2021;20(3):825-834.
    PMID 32799362 · doi:10.1111/jocd.13676 · PMC8176521
24. Figueres Juher T, Basés Pérez E. [An overview of the beneficial effects of hydrolysed collagen intake on joint and bone health and on skin ageing]. Nutr Hosp. 2015;32 Suppl 1:62-6.
    PMID 26267777 · doi:10.3305/nh.2015.32.sup1.9482
25. García-Coronado JM, Martínez-Olvera L, Elizondo-Omaña RE, Acosta-Olivo CA, Vilchez-Cavazos F, Simental-Mendía LE, et al.. Effect of collagen supplementation on osteoarthritis symptoms: a meta-analysis of randomized placebo-controlled trials. Int Orthop. 2019;43(3):531-538.
    PMID 30368550 · doi:10.1007/s00264-018-4211-5
26. Gelse K, Pöschl E, Aigner T. Collagens--structure, function, and biosynthesis. Adv Drug Deliv Rev. 2003;55(12):1531-46.
    PMID 14623400 · doi:10.1016/j.addr.2003.08.002
27. Gorissen SHM, Crombag JJR, Senden JMG, Waterval WAH, Bierau J, Verdijk LB, et al.. Protein content and amino acid composition of commercially available plant-based protein isolates. Amino Acids. 2018;50(12):1685-1695.
    PMID 30167963 · doi:10.1007/s00726-018-2640-5 · PMC6245118
28. Heinemeier KM, Schjerling P, Heinemeier J, Magnusson SP, Kjaer M. Lack of tissue renewal in human adult Achilles tendon is revealed by nuclear bomb (14)C. FASEB J. 2013;27(5):2074-9.
    PMID 23401563 · doi:10.1096/fj.12-225599 · PMC3633810
29. Heinemeier KM, Schjerling P, Øhlenschlæger TF, Eismark C, Olsen J, Kjær M. Carbon-14 bomb pulse dating shows that tendinopathy is preceded by years of abnormally high collagen turnover. FASEB J. 2018;32(9):4763-4775.
    PMID 29570396 · doi:10.1096/fj.201701569R
30. Hexsel D, Zague V, Schunck M, Siega C, Camozzato FO, Oesser S. Oral supplementation with specific bioactive collagen peptides improves nail growth and reduces symptoms of brittle nails. J Cosmet Dermatol. 2017;16(4):520-526.
    PMID 28786550 · doi:10.1111/jocd.12393
31. Honvo G, Lengelé L, Charles A, Reginster JY, Bruyère O. Role of Collagen Derivatives in Osteoarthritis and Cartilage Repair: A Systematic Scoping Review With Evidence Mapping. Rheumatol Ther. 2020;7(4):703-740.
    PMID 33068290 · doi:10.1007/s40744-020-00240-5 · PMC7695755
32. Inoue N, Sugihara F, Wang X. Ingestion of bioactive collagen hydrolysates enhance facial skin moisture and elasticity and reduce facial ageing signs in a randomised double-blind placebo-controlled clinical study. J Sci Food Agric. 2016;96(12):4077-81.
    PMID 26840887 · doi:10.1002/jsfa.7606
33. Iwai K, Hasegawa T, Taguchi Y, Morimatsu F, Sato K, Nakamura Y, et al.. Identification of food-derived collagen peptides in human blood after oral ingestion of gelatin hydrolysates. J Agric Food Chem. 2005;53(16):6531-6.
    PMID 16076145 · doi:10.1021/jf050206p
34. Jacinto JL, Nunes JP, Gorissen SHM, Capel DMG, Bernardes AG, Ribeiro AS, et al.. Whey Protein Supplementation Is Superior to Leucine-Matched Collagen Peptides to Increase Muscle Thickness During a 10-Week Resistance Training Program in Untrained Young Adults. Int J Sport Nutr Exerc Metab. 2022;32(3):133-143.
    PMID 35042187 · doi:10.1123/ijsnem.2021-0265
35. Jendricke P, Centner C, Zdzieblik D, Gollhofer A, König D. Specific Collagen Peptides in Combination with Resistance Training Improve Body Composition and Regional Muscle Strength in Premenopausal Women: A Randomized Controlled Trial. Nutrients. 2019;11(4).
    PMID 31010031 · doi:10.3390/nu11040892 · PMC6521629
36. Jerger S, Centner C, Lauber B, Seynnes O, Sohnius T, Jendricke P, et al.. Effects of specific collagen peptide supplementation combined with resistance training on Achilles tendon properties. Scand J Med Sci Sports. 2022;32(7):1131-1141.
    PMID 35403756 · doi:10.1111/sms.14164
37. Jerger S, Centner C, Lauber B, Seynnes O, Friedrich T, Lolli D, et al.. Specific collagen peptides increase adaptions of patellar tendon morphology following 14-weeks of high-load resistance training: A randomized-controlled trial. Eur J Sport Sci. 2023;23(12):2329-2339.
    PMID 37424319 · doi:10.1080/17461391.2023.2232758
38. Jerger S, Jendricke P, Centner C, Bischof K, Kohl J, Keller S, et al.. Effects of Specific Bioactive Collagen Peptides in Combination with Concurrent Training on Running Performance and Indicators of Endurance Capacity in Men: A Randomized Controlled Trial. Sports Med Open. 2023;9(1):103.
    PMID 37935999 · doi:10.1186/s40798-023-00654-9 · PMC10630299
39. Kadler KE, Baldock C, Bella J, Boot-Handford RP. Collagens at a glance. J Cell Sci. 2007;120(Pt 12):1955-8.
    PMID 17550969 · doi:10.1242/jcs.03453
40. Kadler KE, Hill A, Canty-Laird EG. Collagen fibrillogenesis: fibronectin, integrins, and minor collagens as organizers and nucleators. Curr Opin Cell Biol. 2008;20(5):495-501.
    PMID 18640274 · doi:10.1016/j.ceb.2008.06.008 · PMC2577133
41. Khatri M, Naughton RJ, Clifford T, Harper LD, Corr L. The effects of collagen peptide supplementation on body composition, collagen synthesis, and recovery from joint injury and exercise: a systematic review. Amino Acids. 2021;53(10):1493-1506.
    PMID 34491424 · doi:10.1007/s00726-021-03072-x · PMC8521576
42. Kim DU, Chung HC, Choi J, Sakai Y, Lee BY. Oral Intake of Low-Molecular-Weight Collagen Peptide Improves Hydration, Elasticity, and Wrinkling in Human Skin: A Randomized, Double-Blind, Placebo-Controlled Study. Nutrients. 2018;10(7).
    PMID 29949889 · doi:10.3390/nu10070826 · PMC6073484
43. Kirmse M, Oertzen-Hagemann V, de Marées M, Bloch W, Platen P. Prolonged Collagen Peptide Supplementation and Resistance Exercise Training Affects Body Composition in Recreationally Active Men. Nutrients. 2019;11(5).
    PMID 31126103 · doi:10.3390/nu11051154 · PMC6566878
44. Kirmse M, Lottmann TM, Volk NR, DE Marées M, Holwerda AM, VAN Loon LJC, et al.. Collagen Peptide Supplementation during Training Does Not Further Increase Connective Tissue Protein Synthesis Rates. Med Sci Sports Exerc. 2024;56(12):2296-2304.
    PMID 39086044 · doi:10.1249/MSS.0000000000003519
45. König D, Oesser S, Scharla S, Zdzieblik D, Gollhofer A. Specific Collagen Peptides Improve Bone Mineral Density and Bone Markers in Postmenopausal Women-A Randomized Controlled Study. Nutrients. 2018;10(1).
    PMID 29337906 · doi:10.3390/nu10010097 · PMC5793325
46. Lis DM, Baar K. Effects of Different Vitamin C-Enriched Collagen Derivatives on Collagen Synthesis. Int J Sport Nutr Exerc Metab. 2019;29(5):526-531.
    PMID 30859848 · doi:10.1123/ijsnem.2018-0385
47. Liu X, Machado GC, Eyles JP, Ravi V, Hunter DJ. Dietary supplements for treating osteoarthritis: a systematic review and meta-analysis. Br J Sports Med. 2018;52(3):167-175.
    PMID 29018060 · doi:10.1136/bjsports-2016-097333
48. Lugo JP, Saiyed ZM, Lau FC, Molina JP, Pakdaman MN, Shamie AN, et al.. Undenatured type II collagen (UC-II®) for joint support: a randomized, double-blind, placebo-controlled study in healthy volunteers. J Int Soc Sports Nutr. 2013;10(1):48.
    PMID 24153020 · doi:10.1186/1550-2783-10-48 · PMC4015808
49. Lugo JP, Saiyed ZM, Lane NE. Efficacy and tolerability of an undenatured type II collagen supplement in modulating knee osteoarthritis symptoms: a multicenter randomized, double-blind, placebo-controlled study. Nutr J. 2016;15:14.
    PMID 26822714 · doi:10.1186/s12937-016-0130-8 · PMC4731911
50. Martínez-Puig D, Costa-Larrión E, Rubio-Rodríguez N, Gálvez-Martín P. Collagen Supplementation for Joint Health: The Link between Composition and Scientific Knowledge. Nutrients. 2023;15(6).
    PMID 36986062 · doi:10.3390/nu15061332 · PMC10058045
51. McAlindon TE, Nuite M, Krishnan N, Ruthazer R, Price LL, Burstein D, et al.. Change in knee osteoarthritis cartilage detected by delayed gadolinium enhanced magnetic resonance imaging following treatment with collagen hydrolysate: a pilot randomized controlled trial. Osteoarthritis Cartilage. 2011;19(4):399-405.
    PMID 21251991 · doi:10.1016/j.joca.2011.01.001
52. Miyamoto N, Ishihara K, Oshima T, Kawai M, Oritani Y, Iemoto N. Collagen Peptide Supplementation Enhances Muscle-Tendon Stiffness and Explosive Strength: A 16-wk Randomized Controlled Trial. Med Sci Sports Exerc. 2025;57(12):2877-2886.
    PMID 40623147 · doi:10.1249/MSS.0000000000003814
53. Moskowitz RW. Role of collagen hydrolysate in bone and joint disease. Semin Arthritis Rheum. 2000;30(2):87-99.
    PMID 11071580 · doi:10.1053/sarh.2000.9622
54. Myllyharju J, Kivirikko KI. Collagens, modifying enzymes and their mutations in humans, flies and worms. Trends Genet. 2004;20(1):33-43.
    PMID 14698617 · doi:10.1016/j.tig.2003.11.004
55. Möller I, Martínez K, Terradillos-Guillén A, Costa-Larrión E, Martínez-Puig D, Velasco-Álvarez J. Efficacy and tolerability of native (undenatured) type II collagen supplementation for joint health in healthy volunteers: a randomized double-blind placebo-controlled study. Nutr J. 2026;25(1).
    PMID 41787523 · doi:10.1186/s12937-026-01302-0 · PMC13077837
56. Oertzen-Hagemann V, Kirmse M, Eggers B, Pfeiffer K, Marcus K, de Marées M, et al.. Effects of 12 Weeks of Hypertrophy Resistance Exercise Training Combined with Collagen Peptide Supplementation on the Skeletal Muscle Proteome in Recreationally Active Men. Nutrients. 2019;11(5).
    PMID 31091754 · doi:10.3390/nu11051072 · PMC6566884
57. Oesser S, Adam M, Babel W, Seifert J. Oral administration of (14)C labeled gelatin hydrolysate leads to an accumulation of radioactivity in cartilage of mice (C57/BL). J Nutr. 1999;129(10):1891-5.
    PMID 10498764 · doi:10.1093/jn/129.10.1891
58. Ohara H, Ichikawa S, Matsumoto H, Akiyama M, Fujimoto N, Kobayashi T, et al.. Collagen-derived dipeptide, proline-hydroxyproline, stimulates cell proliferation and hyaluronic acid synthesis in cultured human dermal fibroblasts. J Dermatol. 2010;37(4):330-8.
    PMID 20507402 · doi:10.1111/j.1346-8138.2010.00827.x
59. Oikawa SY, McGlory C, D'Souza LK, Morgan AK, Saddler NI, Baker SK, et al.. A randomized controlled trial of the impact of protein supplementation on leg lean mass and integrated muscle protein synthesis during inactivity and energy restriction in older persons. Am J Clin Nutr. 2018;108(5):1060-1068.
    PMID 30289425 · doi:10.1093/ajcn/nqy193
60. Oikawa SY, Macinnis MJ, Tripp TR, McGlory C, Baker SK, Phillips SM. Lactalbumin, Not Collagen, Augments Muscle Protein Synthesis with Aerobic Exercise. Med Sci Sports Exerc. 2020;52(6):1394-1403.
    PMID 31895298 · doi:10.1249/MSS.0000000000002253
61. Oikawa SY, Kamal MJ, Webb EK, McGlory C, Baker SK, Phillips SM. Whey protein but not collagen peptides stimulate acute and longer-term muscle protein synthesis with and without resistance exercise in healthy older women: a randomized controlled trial. Am J Clin Nutr. 2020;111(3):708-718.
    PMID 31919527 · doi:10.1093/ajcn/nqz332 · PMC7049534
62. Oyamada I, Bird TA, Peterkofsky B. Decreased extracellular matrix production in scurvy involves a humoral factor other than ascorbate. Biochem Biophys Res Commun. 1988;152(3):1490-6.
    PMID 3377781 · doi:10.1016/s0006-291x(88)80454-6
63. Peterkofsky B. Ascorbate requirement for hydroxylation and secretion of procollagen: relationship to inhibition of collagen synthesis in scurvy. Am J Clin Nutr. 1991;54(6 Suppl):1135S-1140S.
    PMID 1720597 · doi:10.1093/ajcn/54.6.1135s
64. Phillips SM, Van Loon LJ. Dietary protein for athletes: from requirements to optimum adaptation. J Sports Sci. 2011;29 Suppl 1:S29-38.
    PMID 22150425 · doi:10.1080/02640414.2011.619204
65. Phillips SM, Tipton KD, van Loon LJ, Verdijk LB, Paddon-Jones D, Close GL. Exceptional body composition changes attributed to collagen peptide supplementation and resistance training in older sarcopenic men. Br J Nutr. 2016;116(3):569-70.
    PMID 27267663 · doi:10.1017/S000711451600221X
66. Praet SFE, Purdam CR, Welvaert M, Vlahovich N, Lovell G, Burke LM, et al.. Oral Supplementation of Specific Collagen Peptides Combined with Calf-Strengthening Exercises Enhances Function and Reduces Pain in Achilles Tendinopathy Patients. Nutrients. 2019;11(1).
    PMID 30609761 · doi:10.3390/nu11010076 · PMC6356409
67. Prockop DJ, Kivirikko KI. Collagens: molecular biology, diseases, and potentials for therapy. Annu Rev Biochem. 1995;64:403-34.
    PMID 7574488 · doi:10.1146/annurev.bi.64.070195.002155
68. Proksch E, Segger D, Degwert J, Schunck M, Zague V, Oesser S. Oral supplementation of specific collagen peptides has beneficial effects on human skin physiology: a double-blind, placebo-controlled study. Skin Pharmacol Physiol. 2014;27(1):47-55.
    PMID 23949208 · doi:10.1159/000351376
69. Proksch E, Schunck M, Zague V, Segger D, Degwert J, Oesser S. Oral intake of specific bioactive collagen peptides reduces skin wrinkles and increases dermal matrix synthesis. Skin Pharmacol Physiol. 2014;27(3):113-9.
    PMID 24401291 · doi:10.1159/000355523
70. Pu SY, Huang YL, Pu CM, Kang YN, Hoang KD, Chen KH, et al.. Effects of Oral Collagen for Skin Anti-Aging: A Systematic Review and Meta-Analysis. Nutrients. 2023;15(9).
    PMID 37432180 · doi:10.3390/nu15092080 · PMC10180699
71. RAMACHANDRAN GN, KARTHA G. Structure of collagen. Nature. 1954;174(4423):269-70.
    PMID 13185286 · doi:10.1038/174269c0
72. RAMACHANDRAN GN, KARTHA G. Structure of collagen. Nature. 1955;176(4482):593-5.
    PMID 13265783 · doi:10.1038/176593a0
73. Ricard-Blum S. The collagen family. Cold Spring Harb Perspect Biol. 2011;3(1):a004978.
    PMID 21421911 · doi:10.1101/cshperspect.a004978 · PMC3003457
74. RICH A, CRICK FH. The molecular structure of collagen. J Mol Biol. 1961;3:483-506.
    PMID 14491907 · doi:10.1016/s0022-2836(61)80016-8
75. Robberechts R, Poffé C, Ampe N, Bogaerts S, Hespel P. Partly Substituting Whey for Collagen Peptide Supplementation Improves Neither Indices of Muscle Damage Nor Recovery of Functional Capacity During Eccentric Exercise Training in Fit Males. Int J Sport Nutr Exerc Metab. 2024;34(2):69-78.
    PMID 37922892 · doi:10.1123/ijsnem.2023-0070
76. Rustad AM, Nickles MA, McKenney JE, Bilimoria SN, Lio PA. Myths and media in oral collagen supplementation for the skin, nails, and hair: A review. J Cosmet Dermatol. 2022;21(2):438-443.
    PMID 34694676 · doi:10.1111/jocd.14567
77. Schadow S, Siebert HC, Lochnit G, Kordelle J, Rickert M, Steinmeyer J. Collagen metabolism of human osteoarthritic articular cartilage as modulated by bovine collagen hydrolysates. PLoS One. 2013;8(1):e53955.
    PMID 23342047 · doi:10.1371/journal.pone.0053955 · PMC3546930
78. Schadow S, Simons VS, Lochnit G, Kordelle J, Gazova Z, Siebert HC, et al.. Metabolic Response of Human Osteoarthritic Cartilage to Biochemically Characterized Collagen Hydrolysates. Int J Mol Sci. 2017;18(1).
    PMID 28117674 · doi:10.3390/ijms18010207 · PMC5297837
79. Schauss AG, Stenehjem J, Park J, Endres JR, Clewell A. Effect of the novel low molecular weight hydrolyzed chicken sternal cartilage extract, BioCell Collagen, on improving osteoarthritis-related symptoms: a randomized, double-blind, placebo-controlled trial. J Agric Food Chem. 2012;60(16):4096-101.
    PMID 22486722 · doi:10.1021/jf205295u
80. Shaw G, Lee-Barthel A, Ross ML, Wang B, Baar K. Vitamin C-enriched gelatin supplementation before intermittent activity augments collagen synthesis. Am J Clin Nutr. 2017;105(1):136-143.
    PMID 27852613 · doi:10.3945/ajcn.116.138594 · PMC5183725
81. Shigemura Y, Iwai K, Morimatsu F, Iwamoto T, Mori T, Oda C, et al.. Effect of Prolyl-hydroxyproline (Pro-Hyp), a food-derived collagen peptide in human blood, on growth of fibroblasts from mouse skin. J Agric Food Chem. 2009;57(2):444-9.
    PMID 19128041 · doi:10.1021/jf802785h
82. Shoulders MD, Raines RT. Collagen structure and stability. Annu Rev Biochem. 2009;78:929-58.
    PMID 19344236 · doi:10.1146/annurev.biochem.77.032207.120833 · PMC2846778
83. Simons VS, Lochnit G, Wilhelm J, Ishaque B, Rickert M, Steinmeyer J. Comparative Analysis of Peptide Composition and Bioactivity of Different Collagen Hydrolysate Batches on Human Osteoarthritic Synoviocytes. Sci Rep. 2018;8(1):17733.
    PMID 30531866 · doi:10.1038/s41598-018-36046-3 · PMC6286367
84. Sivan SS, Wachtel E, Tsitron E, Sakkee N, van der Ham F, Degroot J, et al.. Collagen turnover in normal and degenerate human intervertebral discs as determined by the racemization of aspartic acid. J Biol Chem. 2008;283(14):8796-801.
    PMID 18250164 · doi:10.1074/jbc.M709885200
85. Sun C, Yang A, Teng F, Xia Y. Efficacy of collagen peptide supplementation on bone and muscle health: a meta-analysis. Front Nutr. 2025;12:1646090.
    PMID 41049371 · doi:10.3389/fnut.2025.1646090 · PMC12488437
86. Thorpe CT, Screen HR. Tendon Structure and Composition. Adv Exp Med Biol. 2016;920:3-10.
    PMID 27535244 · doi:10.1007/978-3-319-33943-6_1
87. Van Vijven JP, Luijsterburg PA, Verhagen AP, van Osch GJ, Kloppenburg M, Bierma-Zeinstra SM. Symptomatic and chondroprotective treatment with collagen derivatives in osteoarthritis: a systematic review. Osteoarthritis Cartilage. 2012;20(8):809-21.
    PMID 22521757 · doi:10.1016/j.joca.2012.04.008
88. Verzijl N, DeGroot J, Thorpe SR, Bank RA, Shaw JN, Lyons TJ, et al.. Effect of collagen turnover on the accumulation of advanced glycation end products. J Biol Chem. 2000;275(50):39027-31.
    PMID 10976109 · doi:10.1074/jbc.M006700200
89. Wolfe RR, Church DD, Ferrando AA, Moughan PJ. Consideration of the role of protein quality in determining dietary protein recommendations. Front Nutr. 2024;11:1389664.
    PMID 39606577 · doi:10.3389/fnut.2024.1389664 · PMC11598328
90. Yazaki M, Ito Y, Yamada M, Goulas S, Teramoto S, Nakaya MA, et al.. Oral Ingestion of Collagen Hydrolysate Leads to the Transportation of Highly Concentrated Gly-Pro-Hyp and Its Hydrolyzed Form of Pro-Hyp into the Bloodstream and Skin. J Agric Food Chem. 2017;65(11):2315-2322.
    PMID 28244315 · doi:10.1021/acs.jafc.6b05679
91. Zague V. A new view concerning the effects of collagen hydrolysate intake on skin properties. Arch Dermatol Res. 2008;300(9):479-83.
    PMID 18784933 · doi:10.1007/s00403-008-0888-4
92. Zdzieblik D, Oesser S, Baumstark MW, Gollhofer A, König D. Collagen peptide supplementation in combination with resistance training improves body composition and increases muscle strength in elderly sarcopenic men: a randomised controlled trial. Br J Nutr. 2015;114(8):1237-45.
    PMID 26353786 · doi:10.1017/S0007114515002810 · PMC4594048
93. Zdzieblik D, Oesser S, Gollhofer A, König D. Improvement of activity-related knee joint discomfort following supplementation of specific collagen peptides. Appl Physiol Nutr Metab. 2017;42(6):588-595.
    PMID 28177710 · doi:10.1139/apnm-2016-0390
94. Zdzieblik D, Brame J, Oesser S, Gollhofer A, König D. The Influence of Specific Bioactive Collagen Peptides on Knee Joint Discomfort in Young Physically Active Adults: A Randomized Controlled Trial. Nutrients. 2021;13(2).
    PMID 33562729 · doi:10.3390/nu13020523 · PMC7915677
95. Zdzieblik D, Jendricke P, Oesser S, Gollhofer A, König D. The Influence of Specific Bioactive Collagen Peptides on Body Composition and Muscle Strength in Middle-Aged, Untrained Men: A Randomized Controlled Trial. Int J Environ Res Public Health. 2021;18(9).
    PMID 33946565 · doi:10.3390/ijerph18094837 · PMC8125453
96. Zdzieblik D, Oesser S, König D. Specific Bioactive Collagen Peptides in Osteopenia and Osteoporosis: Long-Term Observation in Postmenopausal Women. J Bone Metab. 2021;28(3):207-213.
    PMID 34520654 · doi:10.11005/jbm.2021.28.3.207 · PMC8441532
97. Food and Agriculture Organization of the United Nations. Dietary protein quality evaluation in human nutrition. Report of an FAO Expert Consultation. FAO Food and Nutrition Paper 92. Rome: FAO; 2013. DIAAS method and the indispensable-amino-acid scoring pattern, including tryptophan.
    Source
98. U.S. Food and Drug Administration. Generally Recognized as Safe (GRAS) notices for gelatin and collagen hydrolysate preparations used as food ingredients. Read as food-ingredient status, not as an efficacy finding.
    Source
99. European Food Safety Authority. Scientific Opinion on the substantiation of a health claim related to collagen hydrolysate and maintenance of joints. EFSA Journal. Read as a claim opinion, not as a trial.
    Source
100. GELITA AG. Product literature distinguishing VERISOL, FORTIGEL, FORTIBONE, BODYBALANCE, and TENDOFORTE as separately specified bioactive collagen peptide compositions. Used only to identify which branded preparation a trial administered, never as evidence of effect.
     Source
101. Rousselot. Peptan collagen peptides technical documentation. Used only to identify the branded hydrolysate named in cited trials.
     Source
102. Lonza / InterHealth. UC-II undenatured type II collagen product identity. Cited to keep this glycosylated native type-II preparation out of the hydrolyzed-peptide evidence set.
     Source
103. ClinicalTrials.gov. Search snapshot for collagen peptide, collagen hydrolysate, VERISOL, FORTIGEL, BODYBALANCE, and UC-II registrations. Identifiers are reported only when a cited paper or the registry record itself names them.
     Source
104. UniProt. Collagen type I alpha-1 chain (COL1A1, P02452) and related human collagen entries. Used for residue composition and domain architecture, not for supplement claims.
     Source

29Evidence handling

Study type is named in the reporting sentence. Animal and cell findings identify a helix, an enzyme, a circulating peptide, or a culture response; they are not used to imply a human outcome. Conflicting trials are left in conflict. A newer null does not automatically supersede an older positive, and an older positive does not automatically survive a larger later trial or an independent tracer study. Named branded mixes are not generalised to generic hydrolysates. Undenatured type II collagen is kept out of the peptide evidence set. Regulator and FAO instruments are cited as instruments. Historical composition work before a usable abstract (Eastoe, 1955) is cited for existence, not for invented residue percentages.

References below were generated from NCBI records fetched for this build. Author lists, titles, journals, years, volumes and identifiers were read from those records. Non-PubMed sources are limited to the FAO 2013 protein-quality consultation, the NIH ODS exercise fact sheet, GRAS/identity pages used only to name products, and manufacturer literature used only to identify which mix a trial administered.

Quantitative claims in the running text are traceable to the project evidence packet. Numeric collagen DIAAS, a printed tryptophan-zero, a Kumar BioCell PMID, a Hijikata Achilles trial, and a percentage of industry dollars are marked unverified and are not treated as findings.

Author–year citations in the prose point at the NCBI records listed below. A paper that names VERISOL, TENDOFORTE, ELASTEN, CH-Alpha, BioCell, UC-II, or “specific collagen peptides” is cited as that preparation, not as the class. Independent tracer trials that used collagen as a low-quality control are cited as protein-quality evidence, not as failed beauty trials. Tables A–F do not introduce numbers that the running text has not already argued.

Where a systematic review pools branded mixes, blends, and generic hydrolysates, the review is reported as a review of that file. It is not promoted into a test of an unassayed grocery powder. Where an abstract omits a conflict statement, the paper is labelled unverified for sponsorship rather than cleared or convicted. That detection limit is itself a finding.

This title is a research review in the South Beach Longevity Science Monograph Series. It is adjacent to, and not a substitute for, the amino-acid nutrition, high-protein diet, hypertrophy, sports-nutrition, and glycine titles. Therapeutic-peptide articles that mention collagen only as a tissue or as a culture substrate are not this document. Injectable medical-device collagen, topical cosmetics as primary evidence, and individualized clinical advice are out of scope.

Nothing in the matrices, the adversarial review, or the running text is a recommendation to buy, avoid, dose, or schedule any preparation. A reader who wants a decision about a person still needs a clinician and a product that was actually tested. This article keeps the four materials, and the three failed class claims, from collapsing into one scoop.

30How a later paper would be read here

A later skin, joint, tendon, bone, or hypertrophy trial is admitted as a test of the preparation it administered, at the dose and duration it used, against the comparator it named. It becomes a class result only if the paper reports the hydrolysate’s source species, mean or modal molecular weight, and peptide composition, and if an independent laboratory repeats the endpoint on a second lot. A trial that reports “collagen peptides” without those identifiers is filed with the generic row, not with VERISOL, TENDOFORTE, FORTIGEL, or any other named mix.

A later protein-quality or tracer study that finds collagen equal to whey, milk, or egg for myofibrillar synthesis, or for lean-mass gain when total essential amino acids are matched, would reopen the muscle-building claim. Until that paper exists, the Oikawa, Jacinto, and Robberechts file, and the Phillips letter beside Zdzieblik 2015, remain the controlling protein evidence. A later tendon trial that is not from the same laboratory as Praet and Jerger, that uses a generic hydrolysate rather than TENDOFORTE, and that reports a clinical rather than a cross-sectional-area endpoint, would be the first independent test of the exercise-plus-peptide care claim. Absence of that paper is recorded as absence, not as a negative proof.

A later skin trial changes the class sentence only if it enrols a grocery hydrolysate whose peptide map is published, uses a clinical rather than a device-only primary endpoint, and is run by a group that does not own the powder. A later joint trial changes the osteoarthritis sentence only if it reports a structural endpoint, names the hydrolysate, and keeps undenatured type II in a separate column. A later bone trial changes the fracture sentence only if it reports fractures, not T-score tenths, and does not co-administer calcium and vitamin D as an unseparated blend. Industry sponsorship, when disclosed, is recorded in the ledger; when the abstract is silent, the paper stays unverified for money rather than being counted as independent.

Sibling titles in this series cover amino-acid nutrition, high-protein diets, hypertrophy, sports nutrition, and glycine as a free amino acid. Those documents are not this one. A reader who wants a whey comparison should read the protein-quality section here and then the hypertrophy title, not a brochure that treats collagen as interchangeable with either. Injectable medical-device collagen, topical cosmetics as primary evidence, and individual clinical advice remain out of scope. The four materials, and the three failed class claims, stay separate after the last page as they were on the first.

Table F. Adversarial resolution

Table F. The six questions this document was written to answer. Detail is in section 24.