The collagen boosters are said to make — type I and type III, and how it is measured
The sentence “type I collagen increases” comes up often in descriptions of boosters. Follow that sentence back to where it came from and most of the time it lands on the colour in a biopsy photograph. But the staining method that produces that colour cannot distinguish collagen types, as validation experiments have reported. This is not a piece written to run down any product; it is a piece that goes class by class to establish what the words “collagen increased” are actually measuring, and how.
The conclusion, first
For no class does there certainly exist a study that shows “type I increased” in human tissue, by a type-specific method, quantitatively, with a control group. The most frequently cited evidence, the colour in a biopsy photograph, is produced by a stain called picrosirius red, and in validation experiments turning the microscope stage by 90 degrees turned red fibres green and green fibres red. Not one thing about the composition had changed, and yet the colour flipped. On top of that, “type III colour” appeared even in the skin of patients with a genetic disease in which type III is deficient. So that colour looks not at the type but at how thick and how densely packed the fibres are and which way they run. In fact the representative human tissue study on CaHA reports that at the treated site the “type I” colour fell from 40.6% to 15.5% and the “type III” colour rose from 1.8% to 34.4% — while other studies in the same class report the opposite direction. The reason the answers diverge is not the product but when you measure and what you measure with. And the widely spread framing that “type III is scar collagen” is not supported in the primary literature — hypertrophic scars have about twice as much type I in absolute terms as normal skin.
What type I · type III · type IV each are
| Type | Where | What |
|---|---|---|
| Type I | Most of the dermis | Thick fibre bundles, tensile strength |
| Type III | At every level of the dermis, alongside type I | Fine fibres, extensibility · elastic recoil |
| Type IV | The basement membrane (the epidermal-dermal junction, vascular basement membranes) | The structure of the interface. It is not in the body of the dermis |
| Type V | Inside type I fibrils | Determines the diameter of type I fibres |
| Type VII | Between the basement membrane and the papillary dermis | Anchoring fibrils — they tie the two layers together |
First, one common misunderstanding. A lot of material describes type III as “young collagen found only in the superficial layer”, but that type I and type III are present together at all levels of the dermis in human skin was already reported in the 1970s. And what sets the diameter of type I fibres is type V. That is, the dichotomy of “type I against type III” is already crude biologically.
The I:III ratio of normal skin differs between sources — as far as we have confirmed it runs from 0.95 if fetal skin is included, and roughly 2.3 to 3 from adolescence onward, diverging with the measurement method (high-performance liquid chromatography / immunohistochemistry / hydroxyproline assay) and with the body site. There is no single “normal value”. You will often see numbers like “young skin is 80% type I · 15% type III”, but we did not reach the primary source for those figures.
The framing that “type III is scar collagen”
The explanation that “you make the good collagen (type I) and avoid the bad collagen (type III)” is widely used. Follow the framing back to where it came from and it lands on the time sequence of wound healing — the observation that type III increases in early granulation tissue and is later replaced by type I. That is a description of “early/late” and not of good and bad.
Measure actual scar tissue and the direction is the reverse.
| Age group | Scar I:III | Normal I:III |
|---|---|---|
| Adolescent | 5.85 | 2.27 |
| Adult | 3.80 | 2.46 |
| Elderly | 2.67 | 2.97 |
The elderly row of this table runs the other way, though — the scar figure of 2.67 is lower than the normal figure of 2.97. In adolescents and adults the I:III ratio of the scar is higher than normal, but in the elderly it was not. We did not find material explaining why, and we leave the table as it is and write this fact down.
In the same material the absolute amount of type I in adolescent scars was about twice that of normal skin. The values another review sets out are also I/III 2.1–3.1 in healthy skin against about 5:1 in scar tissue. That is, what is close to a scar is not an excess of type III but an excess of type I.
The animal work goes a step further. In mice deficient in type III collagen, myofibroblasts increased and the scar area at day 21 after injury was significantly larger than in normal mice. The scarring got worse when type III was lacking. That said, this is mouse material and cannot be carried straight across to a human face. What we can say goes only as far as this: the received idea that “type III is bad” is at least not an established fact in the literature.
To add to that, whether type III increases or decreases with age is another matter on which two primary human studies point in opposite directions. A 1987 study measuring by liquid chromatography said the type III proportion increases in the elderly, while a 2011 study measuring by immunohistochemistry and hydroxyproline said it decreases. The latter appears in a journal whose peer-review reliability is disputed, and we state that alongside.
The method of measurement changes the conclusion
This is the core of the piece. The “type I %” figures for boosters come mostly from photographs stained with picrosirius red and viewed under a polarising microscope. The method measures the area of red and of green.
But that colour does not mean the collagen type. Two things reported by a 2014 validation study are decisive — (1) turning the microscope stage by 90 degrees turned red fibres green and green fibres red. Nothing at all about the composition had changed. (2) Green fibres were observed even in the skin of patients with a genetic disease in which type III collagen is deficient (vascular Ehlers-Danlos). The authors' own sentence runs like this — “Color changes do not reflect any changes in the composition of collagen fibrils.”
In 2021 it was validated again, this time against immunohistochemistry as the reference. The conclusion was “suitable for assessing total collagen but it did not distinguish collagen types”, and it is written that “immunohistochemistry was superior to this stain for detecting type III”.
So what does that colour look at — the thickness of the fibres, the packing density of the bundles, and their orientation. If a procedure rearranges the tissue, the colour can change without a single type changing. And newly made type I fibres, while still fine, also look “green” — which is why the two readings “green = immature” and “green = type III” get mixed together in the material.
The other colour-based stains commonly used on animal tissue are in the same position. People read “blue = new collagen”, but that was inferred from colour, not confirmed with a type-specific antibody.
What has been confirmed in human tissue, class by class
We separate out whether it is human or animal, whether there is a biopsy, and whether the types were distinguished. This is not a table for ranking; it is a table showing how far the evidence has come.
| Class | Human biopsy | Confirmed / empty |
|---|---|---|
| CaHA (Radiesse · DCLASSY) | Yes (n=5–24) | The direction differs from study to study. In immunohistochemistry material at 4 months · 9 months, type I rose from 4.0 to 6.58 and type III fell from 5.2 to 3.7, while in abdominal skin material at 2 months the “type I” colour fell from 40.6% to 15.5% and the “type III” colour rose from 1.8% to 34.4%. In the 24-week material type I went from 64 to 67, within the margin of error |
| PLLA (comparator reference) | Yes | There is a study looking at the human tissue response, but we could not confirm the sample size and quantitative figures at first hand. Another study reports that the locations differ, with type III adjacent to the particles and type I around the capsule, and that the increase in type I is at 8–24 months |
| PCL (GOURI · microsphere) | Case level | The human material that gets cited is either a case with no stated sample size or presented only as a photograph in the body of a review without an author citation. In animals, type III at 9 months → type I at 21 months |
| PDLLA (Juvelook) | We did not find any | The collagen material we confirmed was entirely animal · cell |
| PN (Rejuran) | We did not find any | There are 7 randomised trials · 183 people, but the outcome measures are wrinkles · scars · wound healing and not collagen histology. The collagen evidence is cultured cells. The manufacturer-supported consensus statement itself says that “more histological verification is needed” |
| hADM (Re2O · CellREDM) | There is a clinical trial but no biopsy | The human trial (n=20, 20 weeks) used clinical measures such as skin density · wrinkles · elasticity. In animals (rats) an increase in new collagen density was seen, but type I and type III were not distinguished |
| Type IV · the basement membrane | — | There is human material showing that type IV in the basement membrane decreases with age. But we did not find a human biopsy showing that any booster increased type IV or the basement membrane |
You have to look at the sample sizes alongside. The largest human tissue study in this field is 24 people and most are 5–20. Many compare only before and after the procedure with no control group, and it is common for a study to have a single specimen. One systematic review wrote that every study it covered was rated at “serious risk of bias” for confounding, and the accompanying commentary pointed out that “there is no single-treatment control group at all, so the relative benefit cannot be known”.
A good part of the literature in this field is manufacturer-supported — several of the CaHA-related systematic reviews and reviews we confirmed were manufacturer-sponsored or co-authored by manufacturer employees, and the PN consensus statement was manufacturer-supported too. That does not by itself make them wrong, but you should know it as you read.
The answer changes with when you measure
The single largest reason the conclusions differ from study to study for the same product, the same class, is the time of measurement.
| 2 months (abdomen, PSR) | The “type III” colour increased markedly, the “type I” colour decreased |
|---|---|
| 4 months (immunohistochemistry) | Type III near its peak |
| 9 months (immunohistochemistry) | Type III comes down and type I goes up |
| Animal, 9 months → 21 months | Type III first, then type I |
So the honest answer to “what type of collagen does a booster make?” is “it depends on when you measure”. Measure at 2 months and type III looks to have increased; measure at 9 months and type I looks to have increased. The two results may be not a contradiction but different points in the same process. Marketing material, though, generally cites only the one favourable time point.
You should look at the body site too. The widely cited CaHA tissue evidence and some of the PCL evidence come from abdominal skin (skin excised at abdominoplasty). The abdomen and the face differ in dermal thickness, sebaceous gland · follicle density, mechanical load and degree of photoageing. The principle “do not treat rat dorsal skin as a human face” should carry over into “do not treat human abdominal skin as a human face”.
A gene being switched on and the tissue having changed
In cell experiments you often see results of the form “type I collagen gene expression increased several-fold”. This does not mean that much protein was made. Type I collagen is a gene under strong regulation at the stages after the gene is read, so even when the mRNA rises, protein synthesis · secretion · fibril assembly are regulated separately.
There is a more decisive experiment. PLLA particles failed entirely to stimulate collagen when fibroblasts were cultured alone, and stimulated it only when they were cultured together with macrophages. The title the authors gave it was “Method Is Matter”. The same substance has an effect or has none depending on the experimental design. It also means that the action of a biostimulator comes not from a single cell but from the whole process of the foreign body response.
Collagen having increased and collagen being well arranged
Finally, the most important distinction. Hypertrophic scars and keloids have more collagen. So “collagen up by x%” is in itself neither good news nor bad. What sets the quality of the tissue is not the amount but the arrangement · the thickness · the cross-linking.
There is one variable here that is rarely handled — the myofibroblast. Studies of biostimulators report increases in collagen, but in an animal experiment using PLLA myofibroblast markers increased significantly. Myofibroblasts are the main actors in wound contraction and scarring. “How long the myofibroblasts stay” may matter more to the fate of the tissue than “which type increased”, and we found almost no booster study that reported this item alongside.
Does that mean boosters are useless — no. That things improved clinically is confirmed on a separate axis. There is material from randomised controlled trials in which wrinkle improvement rates came out higher than the control, and we have set those figures out in what has been confirmed for each class of collagen booster. What this piece says is not “there is no effect” but that when the sentence “type I increased by x%” is offered as evidence of clinical effect, the actual weight of that sentence is lighter than it seems.
In summary — what is confirmed, what we could not confirm
| Category | Content |
|---|---|
| Confirmed | Picrosirius red under polarisation does not distinguish collagen types (colour reversal on a 90-degree rotation, “type III colour” observed even in type III-deficient patients, re-validated against immunohistochemistry) · hypertrophic scars have about twice as much type I in absolute terms as normal skin and a higher I:III ratio · type I and type III are present together at all levels of the dermis · type V sets the diameter of type I fibres · PLLA particles failed to stimulate collagen in fibroblast monoculture |
| Could not confirm | Human biopsy collagen type data for PDLLA · PN · hADM · the sample size and quantitative figures of the PCL human biopsy · the quantitative figures of the PLLA human biopsy · human material showing that any booster increased type IV · the basement membrane · the primary source for “young skin 80% type I · 15% type III” |
| Material pointing the other way | The representative CaHA human biopsy reports a decrease in the “type I” colour · scars were larger in mice deficient in type III · the change in type III with ageing runs in opposite directions in two primary human studies · the type I change in the 24-week material is within the margin of error |
| What to know as you read | The largest human tissue study is 24 people · a good part of the evidence is abdominal skin (not the face) · a good part of the literature in this field is manufacturer-sponsored or co-authored by manufacturer employees · one systematic review rated every study it covered at “serious risk of bias” |
The one line this piece is meant to leave. When you see the sentence “type I collagen increases”, ask three things — human or animal, measured with what, measured when. Check those three and the same sentence changes greatly in weight.
Frequently asked questions
Is it true that boosters make type I collagen?
The accurate answer is not “it is untrue” but “a good part of the evidence offered in support of that statement was measured by a method that cannot distinguish collagen types”. For no class did we certainly find a study showing it in human tissue by a type-specific method (immunohistochemistry), quantitatively, with a control group. That things improved clinically is confirmed on a separate axis.
Why can picrosirius red staining not distinguish the types?
Two validation experiments are decisive. Turning the microscope stage by 90 degrees turned red fibres green and green fibres red — not one thing about the composition had changed, and yet the colour flipped. And “type III colour” appeared even in the skin of patients with a genetic disease in which type III is deficient. That colour looks at the thickness · packing density · orientation of the fibres, not at the type.
Is type III collagen not scar collagen?
The primary literature does not support that framing. Hypertrophic scars had about twice as much type I in absolute terms as normal skin, and a higher I:III ratio too (adolescent scar 5.85 against normal 2.27). In animal work, the scar area was actually larger in mice deficient in type III. “Type III early, type I later” describes a time sequence, not good and bad. That said, mouse material cannot be carried straight across to a human face.
Why do the results differ from study to study?
The largest reason is the time of measurement. Measure at 2 months and type III looks to have increased; measure at 9 months and type I looks to have increased. The two results may be not a contradiction but different points in the same process. Add the measurement method (colour-based stain or antibody) and the body site (abdomen or face) on top of that and the conclusions diverge.
Do Rejuran and Juvelook have collagen evidence too?
The state of it differs by class. PN (Rejuran) has 7 randomised trials · 183 people, but the outcome measures are wrinkles · scars · wound healing and not collagen histology — the collagen evidence is at the level of cultured cells, and the manufacturer-supported consensus statement itself says that “more histological verification is needed”. For PDLLA (Juvelook), the collagen material we confirmed was entirely animal · cell.
I have also seen it explained as regenerating type IV collagen.
Type IV is not a component of the body of the dermis but of the basement membrane between the epidermis and the dermis. There is human material showing that type IV in the basement membrane decreases with age. But we did not find a human biopsy showing that any injectable booster increased type IV or the basement membrane. What we confirmed in the related studies was material on topical preparations and three-dimensional skin models, which is a different matter from injectable boosters.
So is more collagen always better?
It is not that simple. Hypertrophic scars and keloids have more collagen. What sets the quality of the tissue is not the amount but the arrangement · the thickness · the cross-linking. And there is one variable that is rarely handled — in an animal experiment using PLLA, myofibroblast markers increased significantly, and myofibroblasts are the main actors in wound contraction and scarring. Almost no booster study reports this item alongside.
What does it mean when a cell experiment says the collagen gene increased several-fold?
It does not mean that much protein was made. Type I collagen is under strong regulation at the stages after the gene is read, so even when the mRNA rises, the actual synthesis · secretion · fibril assembly are regulated separately. More decisive is the experiment in which PLLA particles failed entirely to stimulate collagen when fibroblasts were cultured alone and stimulated it only when cultured together with macrophages. The same substance has an effect or has none depending on the experimental design.
Then why does Miso Clinic do boosters?
This is not a piece saying “there is no effect”. That things improved clinically is confirmed on an axis separate from the biopsy — there is material from randomised controlled trials in which wrinkle improvement rates came out higher than the control, and we have set those figures out separately for each class. What this piece says is that when the sentence “type I increased by x%” is offered as evidence of effect, the weight of that sentence is lighter than it seems.
How large are the studies?
Smaller than you would think. The largest human tissue study in this field is 24 people and most are 5–20. Many compare only before and after the procedure with no control group, and it is common for a study to have a single specimen. One systematic review wrote that every study it covered was rated at “serious risk of bias” for confounding, and the accompanying commentary pointed out that “there is no single-treatment control group at all, so the relative benefit cannot be known”.
Who wrote this
Written and reviewed by Lee Chi-Hak, MD, medical director of Miso Clinic in Daegu, South Korea. Every study cited above is given together with its design, its size and the limitations the authors themselves recorded, and where we could not find data, we have said that we could not find any.
| Medical director | Lee Chi-Hak, MD |
|---|---|
| Address | 4F Bombom Building, 125 Dongdeok-ro, Jung-gu, Daegu, South Korea · Exit 1, Kyungpook National University Hospital Station |
| Phone | +82-53-428-2700 |
| Hours | Weekdays 11:00–19:00 (lunch 13:00–14:00) / Saturday 10:00–16:00 (no lunch break) / Closed Sundays and public holidays |
| Columns | All clinical columns |
| Reference library | All booster and device references |
References
- The limits of the staining method — Lattouf R et al., Journal of Histochemistry & Cytochemistry 2014;62(10):751–758. The polarisation colour reflects the thickness · bundle packing density · orientation of the fibres and not the collagen type. Rotating the microscope stage by 90 degrees swapped red-yellow and green, and green fibres were observed even in the skin of a patient with type III deficiency (Ehlers-Danlos type 4). The authors' original text — “Color changes do not reflect any changes in the composition of collagen fibrils.”
- Re-validation — López De Padilla CM et al., Journal of Histochemistry & Cytochemistry 2021;69(10). Compared against dual immunohistochemistry as the reference. Conclusion — suitable for the qualitative · quantitative assessment of total collagen but it did not distinguish collagen types, with immunohistochemistry superior for detecting type III.
- The I:III of scars — Cheng et al. 2011, 40 human normal skin specimens (10 fetal + 30 from burn patients) and 90 hypertrophic scar specimens, 6–12 months after injury. Adolescent scar I:III 5.85 ± 0.40 against normal 2.27 ± 0.13, adult 3.80 against 2.46, elderly 2.67 against 2.97. Adolescent scar type I 523.12 µg/g against normal 279.12 µg/g. This study appears in a journal whose peer-review reliability is disputed, and we state that alongside. The values a separate review sets out are healthy skin I/III 2.1–3.1 against about 5:1 in scars (a secondary citation).
- Type III deficiency and scarring — Volk SW et al., Cells Tissues Organs 2011;194(1):25–37. A mouse model (Col3-deficient against wild type). Increased wound contraction, increased myofibroblast markers, scar area at day 21 after injury significantly larger than in the wild type. This is animal material and cannot be applied straight to a human face. There is a 2024 follow-up study in the same direction, but we did not confirm its figures.
- Distribution at all levels of the dermis — Epstein EH Jr, “Human skin collagen. Presence of type I and type III at all levels of the dermis.” We did not confirm the year or the original text of the abstract, only the bibliographic details.
- Change with ageing — Lovell CR et al., British Journal of Dermatology 1987;117(4):419–428 (high-performance liquid chromatography · scanning electron microscopy): in covered skin the I:III ratio is constant throughout childhood and young adulthood, with an increased type III proportion in the elderly. The sample size is not stated in the abstract. This is in the opposite direction to Cheng 2011 above.
- CaHA human biopsy — Zerbinati N & Calligaro A, Clinical, Cosmetic and Investigational Dermatology 2018;11:29–35. 5 women, abdominal skin in patients scheduled for abdominoplasty, 2 months after the procedure, picrosirius red + circular polarisation, self-controlled. Red · orange 40.59 ± 6.34% → 15.54 ± 3.21%, green · yellow 1.76 ± 0.48% → 34.42 ± 5.52% (p<0.01). The authors wrote that 2 months may be earlier than the point of maximum type III stimulation (estimated at about 4 months). You should read this together with the point that the measurement method does not distinguish types (the two items above).
- CaHA immunohistochemistry — material on 20 people (4 · 7 months) and 24 people (4 · 9 months). In the latter, type I 4.0 ± 1.44 (4 months) → 6.58 ± 1.1 (9 months), type III 5.2 ± 1.67 → 3.7 ± 1.09. In the 24-week material (13 people), type I 64 ± 12 → 67 ± 14 (within the margin of error), type III 57 ± 11 → 64 ± 10. The units and scales differ from study to study (semi-quantitative score · H-score · fibre count · %), so they cannot be compared with one another, and a good many are secondary citations by way of reviews.
- A systematic review's own assessment — the systematic review of ultrasound in combination + CaHA (Aesthetic Surgery Journal 2025;45(6):638, manufacturer-sponsored). All 11 human studies were rated at “serious risk of bias” for confounding, with the authors' original text — “the histological conclusions are drawn from a single specimen per study, which may weaken their robustness”. The accompanying commentary (same issue, page 643) — “the largest limitation is that there is no single-treatment control group at all, so the relative benefit cannot be known”. Another review of the CaHA mechanism (2025;45(4):393, authors who advise the manufacturer) also concedes that the supporting evidence for quantifying collagen types I/III is lacking.
- The design-dependence of cell experiments — Ray & Ta, Journal of Functional Biomaterials 2020;11(3):51. The title is “Method Is Matter”. PLLA nanoparticles failed to stimulate collagen in fibroblast monoculture, and stimulation was observed only in fibroblast-macrophage co-culture. This directly conflicts with other studies reporting a surge in type I mRNA in cultured cells.
- Myofibroblasts — in an experiment using PLLA in 150 mice (Aesthetic Plastic Surgery 2025), myofibroblast markers increased significantly at the 30-day · 60-day time points (p=0.001, p<0.001). This study did not distinguish type I from type III.
- PN — by the systematic review of randomised trials (Cureus 2025) there are 7 RCTs · 183 people in total, with outcome measures of wrinkles · scars · wound healing. The collagen-related evidence is at the level of cultured fibroblasts, and the expert consensus statement whose writing the manufacturer supported states itself that “more randomised controlled trials, mechanistic studies and histological verification of treatment outcomes are needed”.
- hADM — the human randomised · split-face · double-blind trial (n=20, 20 weeks) looked at clinical measures such as skin density · volume · wrinkle depth · pores · elasticity, and no skin biopsy or collagen type measurement was performed. In the preclinical work (rats, 1 week) new collagen density and fibroblast density increased significantly, but type I and type III were not distinguished.
- PCL — the human material that gets cited is a 13-month · 1-year biopsy of the temple region, but the sample size is not stated, so it appears to be a case. The “abdominal skin, 6-month immunohistochemistry” is presented only as a photograph in the body of a review without an author citation, and we did not find the primary source. In animals (rabbits), type III at 9 months → type I at 21 months.
- PLLA human tissue — there is a single-arm human tissue response study (Dermatologic Surgery 2013), but we could not confirm the sample size · biopsy time points · quantitative figures at first hand. Another study (Journal of Dermatological Science 2015) reports that the locations differ, with type III adjacent to the particles and type I around the capsule, and that the increase in type I is at 8–24 months, but we could not settle whether that study is human tissue or animal.
- Type IV and the basement membrane — there is human material showing that type IV decreases with age at the epidermal-dermal junction and in fibroblasts (European Journal of Dermatology 2016, women aged 30–70), but we could not confirm the quantitative figures. The study showing that basement membrane remodelling affects papillary dermal collagen (Scientific Reports 2022) is material on a topical preparation and a three-dimensional skin model and is a different matter from injectable boosters. We did not find a human biopsy showing that any booster increased type IV.
- The I:III values of normal skin — the widely circulated numbers including “young skin about 80% type I · about 15% type III” are all secondary citations by way of reviews, and we did not reach the primary source. We have not presented them as confirmed values in the body of the piece.
Everything in this column is general information and does not replace medical diagnosis or treatment. Effects and side effects vary with individual skin condition, age and underlying illness, and the same result is not guaranteed for everyone. Any decision to proceed should be made in an in-person consultation with a physician.
← Clinical columns · Reference library · Miso Clinic home
한국어 · English · 日本語 · 简体中文 · Español · Tiếng Việt · ภาษาไทย · Bahasa Indonesia