Only a handful of places left for our September 2026 cohort. Applications close soonApply now →

MSTA
Your future
starts here.

Fibroblast

Also known as: fibroblasts, dermal fibroblast, fibroblast collapse, azficel-T

The fibroblast is the principal synthetic cell of the dermis. It produces collagen, elastin and the extracellular matrix — and also the matrix metalloproteinases that degrade them. Stimulating it is the stated mechanism behind most regenerative skin treatments.

Evidence status

Moderate

The cell biology and the ageing 'mechanical collapse' model rest on decades of primary human in-vivo work, though largely from one research group. Clinical evidence for treatments that claim to stimulate fibroblasts is weaker: dominated by uncontrolled designs, subjective endpoints, and placebo responses of 36–48% where they have been measured.

What the fibroblast does#

The fibroblast is the principal synthetic cell of the dermis. It produces collagen, elastin and the extracellular matrix — and it also produces the matrix metalloproteinases that degrade them.[1, 3]

That second half is routinely left out, and leaving it out misreads the cell. Treating “stimulating collagen” as separate from “MMP output” imagines a cell that only builds. The same cell does both, and the outcome is a balance rather than a total.

What changes with age#

In sun-protected hip skin from six younger and six older participants— so this is intrinsic ageing, not sun damage — tissue type I procollagen was 68% lower in those aged 80 and over than in those aged 18 to 29.[1]

Two things about where that figure sits. It is an in-skin tissue measurement, and it does not appear in the paper’s abstract — an abstract-only check will find a different, smaller in-vitro figure and appear to contradict it.

And tissue content is not per-cell output. Twelve people’s hip skin tells you how much procollagen is in the tissue; how much of that difference is fewer cells, more degradation, a changed matrix or sampling, rather than each fibroblast synthesising less, is not something this experiment separates. The collapse model below is the argument that it is per-cell — an argument, not this measurement.

The collapse model#

The most influential explanation for why aged skin makes less collagen is mechanical rather than metabolic.

Aged fibroblasts were in contact with intact collagen fibrils over 58±8% of their profile, against 78±6% in young skin, and showed roughly half the cell spreading.[1]

The proposed reading: fragmented collagen no longer gives the cell anything to hold onto, the cell collapses, and a collapsed fibroblast under no mechanical tension does not synthesise. Damage becomes self-sustaining.

Two honest caveats

The measurements are two-dimensional, taken from electron-microscope sections rather than true surface area, and the cell counts were unbalanced — 160 young cells against 57 aged.

More importantly, the model itself is set out in a narrative review by the same group that proposed it, whose own abstract frames it as reviewing the state of the art. It is a well-argued hypothesis with real supporting data, not an independently confirmed mechanism.[2]

It is also the strongest available rationale for why controlled dermal injury produces new collagen: the treatment re-establishes a provisional matrix and the mechanical cues the cell has lost. That is an inference from the model, and worth flagging as one.

What UV does to it#

In 59 participants, on buttock skin, a singleUV exposure induced collagenase, a 92-kDa gelatinase and stromelysin — MMP-1, MMP-9 and MMP-3 in modern numbering — and raised degradation of endogenous type I collagen by 58%.[3]

Four exposures at two-day intervals held collagenase at 4.4 times baseline and gelatinase at 2.3 times, for a full seven days.

That is the mechanism behind photoageing stated cleanly: not collagen slowly wearing out, but enzymes being switched on and staying on.

The retinoid finding — and its limits#

Tretinoin pretreatment inhibited UV-induced MMP protein and activity by 70–80%.[3, 4]

This is the single most misused finding in this entry, so the scope needs stating clearly. The tretinoin was applied under occlusion, 48 hours before a single controlled laboratory irradiation.

It does not show that a retinoid protects skin from everyday sun exposure, and it does not make a retinoid a substitute for sunscreen. It shows that the MMP induction pathway is pharmacologically interruptible, which is interesting and different.

What resurfacing actually triggers#

The molecular sequence after ablative resurfacing is striking, and almost always quoted without its scope.

After focal CO2 resurfacing in 28 volunteers, MMP-1 messenger RNA rose roughly 39,000-fold, and type I and III procollagen mRNA peaked at 7.5× and 8.9× baseline at day 21, still elevated at six months.[5]

Four things belong with those numbers:

  • They are messenger RNA — the signal to make the protein, not the protein, and not enzyme activity.
  • It was forearm skin, not face — different thickness, adnexal density and healing behaviour.
  • It was focal treatment of a test site, not a full-face protocol.
  • Single-arm, no control group, and six months is where follow-up ended — so it cannot say when levels returned to baseline.

The shape of the finding is real and useful: demolition first, rebuilding after, over months. The precision implied by “39,000-fold” is not.

Injecting fibroblasts: the placebo arm#

If the premise is that more fibroblasts means better skin, then injecting cultured fibroblasts is the purest available test of it. That trial has been done properly — multicentre, double-blind, placebo-controlled, 372 adults.

At six months, 78% of treated participants rated themselves at least one point improved, against 48% on placebo. On blinded evaluator assessment, 64% versus 36%.[6]

The treatment beat placebo convincingly. But look at the other column.

Between 36% and 48% of people who received nothing recorded improvement.[6]

That is the most useful number on this page. Any uncontrolled before-and-after series — which is to say almost every result you will see marketed — is measuring at least that much non-specific improvement before any treatment effect at all.

What the regulator actually approved it on

The published trial used a threshold of at least onepoint of improvement. The regulator’s co-primary endpoint required two — and those numbers live in the FDA record, not in the journal paper. Two vehicle-controlled trials, 421 subjects aged 23 to 81, assessed six months after the third session:

  • Subject assessment — 57% vs 30% vehicle in Study One; 45% vs 18% in Study Two.
  • Physician assessment — 33% vs 7%; 19% vs 7%.[9]

So 78% versus 48% is not the efficacy the product was licensed on. And note that the vehicle response does not vanish at the stricter threshold: 30% and 18% of people who received vehicle still rated themselves two points better.

The label is unusually candid about the rest. The mechanism is unknown. Efficacy beyond six months is not established. Non-White subjects were 8% of the population, so there is insufficient information to assess efficacy in non-Whites, and geriatric responder rates were lower and less consistent. Manufacturing failed outright for 6.2% of subjects randomised to the product, and a further 5.7% had too little to complete three sessions.[9]

RF microneedling and needling#

A systematic review of radiofrequency microneedling covered 20 studies and 558 patients — but the headline count flatters every outcome inside it. Texture improvement was documented across eight studies and lifting or tightening across five. Adverse events were mild and transient, mostly erythema and oedema settling within hours to days.[7]

The design tells you how much weight to put on it: eight of the 20 were case series, eight were prospective studies, and only four were randomised controlled trials. Study sizes ranged from 9 to 133, averaging about 32. Three of the review’s five authors are employed by or speak for a radiofrequency microneedling device manufacturer whose device features among the included studies — disclosed in the reference list.

For needling in acne scarring, Cochrane pooled 24 trials and 789 adults and found a lack of high-quality evidence overall — poor methodology, underpowered studies, no standardised assessment. On needling specifically, no comparison rose above low-quality evidence.[8]

In professional practice#

  • Remember the cell does both jobs. Stimulating fibroblasts is not the same as net collagen gain.
  • Hold the placebo number in mind. 36–48% improved on nothing. Your uncontrolled results include that.
  • Don’t quote mRNA figures as outcomes. 39,000-fold is a transcript measurement on forearm skin.
  • Never present a retinoid as sun protection.The 70–80% MMP finding was a single controlled irradiation of buttock skin in a laboratory.[3, 4]
  • Be honest about the needling evidence. Cochrane found nothing above low quality. That supports cautious claims and openly stated uncertainty. Whether to offer a procedure is a practice decision with its own governance basis — not something an evidence synthesis can tell you, and not something we will read into one here.

What remains uncertain#

  • Whether the collapse model is causal or a correlate of ageing.
  • Whether transcript-level increases translate into durable measurable collagen in human skin.
  • How much of any aesthetic result is the treatment and how much is the 36–48% non-specific response.
  • Whether radiofrequency microneedling outperforms conventional needling — the trials are too few and too small to say.

Common misconceptions#

“Fibroblasts build collagen — stimulate them and you gain collagen.”

They also produce the enzymes that degrade it. The outcome is a balance.[1, 3]

“Retinoids protect against UV damage.”

Under occlusion, 48 hours before one controlled irradiation. Not everyday sun, and not instead of sunscreen.[3, 4]

“Laser increases collagen 900%.”

The real figures are messenger RNA — the signal, not the protein — measured on forearm skin in a single-arm study.[5]

“The before-and-afters prove it works.”

36–48% of placebo participants recorded improvement in a blinded trial. Uncontrolled photos cannot separate that from treatment.[6]

“RF microneedling is well evidenced.”

20 studies, of which four were randomised, in a review with manufacturer-affiliated authors.[7]

Frequently asked questions#

Why do results vary so much between clients?

Partly technique and partly biology — but also because a large share of perceived improvement is non-specific. In a blinded trial, 36–48% of placebo participants improved.[6]

Does injury really make skin build collagen?

The molecular sequence is well documented — demolition then rebuilding, with procollagen signal peaking around day 21 and still elevated at six months. Those are transcript measurements on forearm skin, so treat the shape as real and the magnitude as uncertain.[5]

Why does aged skin respond differently?

It starts from 68% less type I procollagen, and its fibroblasts have less contact with intact collagen to hold onto.[1]

Is RF microneedling better than standard microneedling?

Not established. The RF review had only four randomised trials among 20 studies, and Cochrane found no needling comparison above low-quality evidence.[8]

References#

Each source is graded by evidence tier. Tier 4 material (manufacturer documents, expert consensus, practitioner experience) is useful for protocol and context, and is never presented as equivalent to independent clinical evidence.

  1. Varani J, Dame MK, Rittié L, Fligiel SEG, Kang S, Fisher GJ, Voorhees JJ. Decreased collagen production in chronologically aged skin. American Journal of Pathology. 2006;168(6):1861–1868.Tier 3Supports: In sun-protected skin, type I procollagen was 68% lower in people aged 80+ than 18–29 — an IN-SKIN tissue measurement that does not appear in the abstract (the abstract reports only in-vitro fibroblast figures, 82±16 vs 56±8 ng/ml). Aged fibroblasts contacted intact collagen fibrils over 58±8% of their profile versus 78±6% in young skin (P<0.01) — a two-dimensional measure from electron-microscope sections, not true surface area. Cell spreading was roughly half, as a normalised index, on unbalanced counts of 160 young versus 57 aged cells.
  2. Fisher GJ, Varani J, Voorhees JJ. Looking older: fibroblast collapse and therapeutic implications. Archives of Dermatology. 2008;144(5):666–672.Tier 4Supports: States the mechanical-collapse model. IMPORTANT: this is a narrative review and stated hypothesis BY THE GROUP THAT PROPOSED THE MODEL — its own abstract frames it as reviewing state-of-the-art knowledge. It is not independent confirmation.
  3. Fisher GJ, Wang ZQ, Datta SC, Varani J, Kang S, Voorhees JJ. Pathophysiology of premature skin aging induced by ultraviolet light. New England Journal of Medicine. 1997;337(20):1419–1428.Tier 2Supports: In 59 white participants aged 21–58, on BUTTOCK skin: a single UV exposure induced collagenase, a 92-kDa gelatinase and stromelysin (MMP-1, MMP-9 and MMP-3 in modern numbering), and degradation of endogenous type I collagen fibrils rose 58%. Four exposures at two-day intervals held collagenase at 4.4-fold and gelatinase at 2.3-fold baseline for seven days.
  4. Fisher GJ, Datta SC, Talwar HS, et al. Molecular basis of sun-induced premature skin ageing and retinoid antagonism. Nature. 1996;379(6563):335–339.Tier 2Supports: Tretinoin pretreatment inhibited UV-induced matrix metalloproteinase protein and activity by 70–80%. CRITICAL SCOPE: the tretinoin was applied UNDER OCCLUSION 48 hours before a single controlled laboratory irradiation. This does not show that a retinoid protects against everyday sun exposure or substitutes for sunscreen.
  5. Orringer JS, Kang S, Johnson TM, et al. Connective tissue remodeling induced by carbon dioxide laser resurfacing of photodamaged human skin. Archives of Dermatology. 2004;140(11):1326–1332.Tier 3Supports: 28 volunteers aged 48–76, single-arm with no control group. FOREARM skin, FOCAL treatment — not full-face resurfacing. Messenger RNA measurements: MMP-1 roughly 39,000-fold, MMP-3 1,041-fold, MMP-13 767-fold, MMP-9 75-fold above baseline. Type I and III procollagen mRNA peaked at 7.5× and 8.9× baseline at day 21 and remained elevated at six months, where follow-up ended. These are transcript-level measurements, not enzyme protein or activity, and no timepoint is stated for the MMP figures.
  6. Smith SR, Munavalli G, Weiss R, Maslowski JM, Hennegan KP, Novak JM. A multicenter, double-blind, placebo-controlled trial of autologous fibroblast therapy for the treatment of nasolabial fold wrinkles. Dermatologic Surgery. 2012;38(7 Pt 2):1234–1243.Tier 2Supports: 372 adults enrolled and treated with autologous cultured fibroblasts (azficel-T) or placebo, three treatments at five-week intervals. At six months, at least a ONE-point wrinkle improvement was recorded by 78% of treated versus 48% of placebo participants on self-assessment, and 64% versus 36% on blinded evaluator assessment (both P<0.001). SCOPE: this publication reports that trial at the one-point threshold. The regulatory endpoint, the two-trial 421-participant programme and the licensing figures belong to the FDA record (source 9) and must not be attributed here.Funding / interest: This is the pivotal trial of a commercial product: azficel-T is Fibrocell's, and the FDA label requires administration only by providers who have completed a Fibrocell-approved training programme. We could NOT retrieve this paper's own funding and conflict-of-interest statement — it sits behind the publisher paywall. Do not present it as declared-clean; treat it as a company-product efficacy publication whose disclosures we have not read.
  7. Kumar N, Suh DH, Lee SJ, Kasif SA, Carruthers JDA. Radiofrequency microneedling for facial rejuvenation: a systematic review. Journal of Cosmetic Dermatology. 2026.Tier 1Supports: 20 studies, 558 patients, published 2015–2025 (the methods paragraph says 22, but the PRISMA diagram, results and quality appraisal all work from 20). Study sizes 9 to 133, mean about 32. Design: eight case series, eight prospective studies, four randomised controlled trials. Texture improvement documented across eight studies; lifting and tightening across five. Adverse events mild and transient.Funding / interest: Three of the five authors are employed by or speak for Jeisys Medical Inc, a radiofrequency microneedling device manufacturer whose INTRAcel device features among the included studies.
  8. Abdel Hay R, Shalaby K, Zaher H, et al. Interventions for acne scars. Cochrane Database of Systematic Reviews. 2016;(4):CD011946.Tier 1Supports: 24 trials, 789 adults. Found a lack of high-quality evidence, attributed to poor methodology, underpowered studies, no standardised assessment of improvement and differing baselines. On needling specifically, no comparison rose above low-quality evidence.
  9. US Food and Drug Administration. LAVIV (azficel-T) suspension for intradermal injection: full prescribing information. Fibrocell Technologies. Revised June 2011.Tier 1Supports: THE REGULATORY PRIMARY RECORD for the pivotal programme — the endpoint and denominators belong here, not to the journal publication. Effectiveness was demonstrated in TWO identically designed, multi-centre, randomised, double-blind, VEHICLE-CONTROLLED studies: 421 subjects aged 23 to 81, randomised to LAVIV (n=210) or vehicle control (n=211), three treatment sessions at three- to six-week intervals. CO-PRIMARY OUTCOMES were the proportion achieving a TWO-point improvement from baseline in nasolabial fold appearance six months after the third session, assessed independently by blinded subjects and blinded evaluating physicians. Results — Study One: subject assessment 57% (57/100) LAVIV vs 30% (31/103) vehicle (p=0.0001); physician assessment 33% (33/100) vs 7% (7/103) (p<0.0001). Study Two: subject 45% (50/110) vs 18% (19/108) (p<0.0001); physician 19% (21/110) vs 7% (8/108) (p=0.0075). LIMITS STATED ON THE LABEL: 'The mechanism by which LAVIV improves the appearance of nasolabial fold wrinkles is unknown'; efficacy beyond six months has not been established; non-White subjects were 8% of the study population, so 'there is insufficient information to assess the efficacy of LAVIV in non-Whites'; geriatric responder rates were 'lower and less consistent'. MANUFACTURING: 6.2% of subjects randomised to LAVIV received none because manufacture failed, and a further 5.7% had insufficient product to complete three sessions.Funding / interest: The product's manufacturer is Fibrocell. Only healthcare providers who have completed a Fibrocell-approved training programme may administer it.

Continue learning with MSTA#

Skinipedia is written by the educators at MSTA, the medic-led skincare training academy in Liverpool.