Collagen types

What are collagen types 1, 2 and 3?

The collagen family consists of 28 members and is numbered with Roman numerals (I–XXVIII). ^[1] Types I, II and III are fibril-forming collagens; fibril-forming collagens are the most abundant collagens in vertebrates and contribute tensile strength in skin and resistance to traction in ligaments. ^[1]

• Type I: The triple helix structure makes up about 96% of the molecule. Fibrils composed of collagen I and III are found in skin. ^[1]
• Type II: In cartilage, collagen II forms fibrils with XI and IX, or with II and III. ^[1]
• Type III: Forms fibrils together with collagen I in skin; it can also be seen together with II in cartilage. ^[1]

This information answers the question “which tissue uses which collagen in the body.” It does not answer the question “which supplement goes to which tissue”; the rest of this guide concerns that second question.

What is the difference between hydrolyzed collagen and undenatured type II collagen?

The two products, although their names are similar, are prepared in different forms and examined with different assumptions.

Hydrolyzed collagen (collagen peptides): Collagen obtained from animal tissue (skin, bone, cartilage) that has been broken down into small peptides by enzymes. In healthy volunteers, after gelatin hydrolysates obtained from pig skin, chicken feet and cartilage, the peptide form of hydroxyproline in plasma peaked at 1–2 hours, and the main peptide was identified as Pro-Hyp. ^[7] Considering that the triple helix is made up of Gly-X-Y repeats, that is, rich in glycine, proline and hydroxyproline, ^[1] it is understood that these peptides are not specific to a single collagen type; type information does not remain a defining feature in a product broken into peptides. This is an inference; no type-based comparative study is included in the evidence reviewed.

Undenatured type II collagen: Whole type II collagen obtained from chicken breast cartilage. In 191 people aged 40–75 with knee osteoarthritis, 40 mg per day of undenatured type II collagen, placebo and glucosamine-chondroitin were compared for 180 days; the total WOMAC score, the primary endpoint of the study, fell more on day 180 in this group than with placebo (p = 0.002) and glucosamine-chondroitin (p = 0.04), and among secondary endpoints the WOMAC pain subscore was also significant versus placebo (p = 0.0003). ^[6] This is the result of a single sponsored randomized study; a finding based on a single study is at the level of limited evidence and cannot be carried over to hydrolyzed collagen. The study was sponsored by a supplement company, and two authors are employees of that company; the authors state that additional studies explaining the mechanism are needed. ^[6]

The mechanism discussed by the authors is the hypothesis that orally taken undenatured type II collagen activates regulatory T cells (Treg); this hypothesis rests on animal and laboratory studies, and the authors report that broken-down (hydrolyzed) type II collagen did not protect against the development of arthritis in animals. ^[6] That is, the result of the undenatured type II collagen study cannot be used for hydrolyzed collagen, and the results of hydrolyzed collagen studies cannot be used for undenatured type II collagen.

Does molecular weight (Da) matter?

The degree of hydrolysis determines the average peptide length and is measured as average molecular weight; however, molecular weight does not indicate the identity of the peptides, which is determined by the amino acid sequence. ^[4] The phrase “low Da” in product promotion is therefore not a quality criterion on its own.

The products used in the studies differ in weight: about 3 kDa on average in the muscle study in older men ^[2], and about 5 kDa in the bone study in postmenopausal women ^[3]. The study comparing directly is a small pilot in healthy volunteers: hydrolysates of fish, porcine and bovine origin (two bovine products of 2,000 and 5,000 Da) were given as a single dose, free hydroxyproline uptake was found to be similar, and total hydroxyproline being higher than the free form showed that di- and tripeptides containing hydroxyproline are absorbed. The authors state that all hydrolysates led to the relevant plasma levels regardless of source and molecular weight, and that larger studies are needed to determine the ideal level. ^[4] The study had 6 participants and was funded by the hydrolysate manufacturer; some of the authors are employees of that company. ^[4] The claim that “lower Da gives a better result” is therefore not supported by clinical outcomes.

Is marine collagen better than bovine collagen?

In the meta-analysis of skin studies, collagen of fish origin was used in 14 of 26 RCTs, bovine in 1, chicken in 1 and porcine in 2, and the source was not stated in 9 studies. The paper reports significant improvement versus placebo in subgroups by source for hydration (fish, bovine, chicken, porcine, unknown source); however, the reported effect of chicken-derived collagen is weakest, and (−0.03; 95% CI −0.40 to 0.34) includes zero. For elasticity, no significant difference was found between sources (p = 0.21). ^[5] Because there were only 1–2 studies in the bovine, chicken and porcine subgroups, the power of this comparison is very limited; the result “fish is better” cannot be drawn from this, and it can only be said that fish is the most studied source. ^[5]

The source animal can also matter with regard to allergens and religious/dietary preferences; these topics are covered on the harms page.

How do type claims in marketing match the evidence?

In the table, the left column shows how type information is used on labels; the middle column the actual distribution in tissue; and the right column what kind of study the supplement was tested with. The three columns cannot be automatically translated into one another: the presence of a type in a tissue does not mean that swallowing that type reaches that tissue.

Are “types 1, 2, 3 together” products more effective?

In the skin, bone and muscle studies reported, hydrolysates were defined not by type mixture but by source animal and processing method; these studies do not compare type combinations. For example, in the muscle study in older men, the product was obtained “by breaking down type I collagen” and was given at the gram level. ^[2] The authors tested a single product in that study; whether type combinations improve or worsen the result therefore cannot be said from this study.

In practice, the questions to ask when reading a label are these: is the product hydrolyzed or undenatured? Is the source animal stated? Does the amount match the gram level in the studies or the milligram level in the undenatured type II collagen study? This guide does not recommend a product or dose.