Evidence levels, and how to read them

“Studies show” can mean a cell-culture experiment or a meta-analysis of randomised trials, and the gap between those is the whole question. In regenerative aesthetics the gap is unusually wide: much of what is marketed rests on mechanism and animal work, while the one modality with real trial evidence — PRP for pattern hair loss — was assessed by its own systematic reviewers as low certainty. Reaching a high tier is necessary but not sufficient; a systematic review of poor studies is still a review of poor studies, and it will usually say so if you read past the abstract.

The ladder, and who is standing on which rung

Each tier answers a different question. Mechanism and cell-culture work answer “could this plausibly do something?” Case series answer “did something happen when we tried it?” Randomised controlled trials answer “did it happen because of the treatment rather than time, placebo, or selection?” Systematic reviews and meta-analyses answer “does that hold across everyone who has tested it?”

Marketing copy routinely borrows the language of the top tier to describe findings from the bottom one. A phrase like “clinically proven to stimulate collagen” is compatible with a fibroblast culture experiment. It is worth training yourself to notice which question a cited study was built to answer.

The case study: PRP for pattern hair loss

PRP in androgenetic alopecia is the most useful example in this field because the research community actually completed the sequence. There are randomised controlled trials. There are multiple independent meta-analyses pooling them. And the conclusions are instructive.

The largest, covering 27 controlled trials and 1,117 subjects, found that PRP increased hair density relative to saline injection over medium-term follow-up — a mean difference of about 25.6 hairs per square centimetre — and then rated that evidence as low quality under GRADE, citing inconsistency and risk of bias. It also noted no serious adverse events attributable to PRP^[1]. A separate meta-analysis restricted to nine randomised controlled trials in 238 patients found increased hair density at three and six months against placebo, while hair count and hair diameter improved from baseline without reaching significance against placebo^[2]. A third, restricted to female pattern hair loss across seven studies, found increased terminal hair density but not thickness, and its authors explicitly recommended caution until the finding is replicated in larger and more representative samples^[3].

Read together those three tell a consistent story: a real but modest signal on one outcome, inconsistent effects on others, and reviewers who decline to overstate it. That is what honest evidence looks like when it is neither nothing nor proof.

Where the other modalities sit

Polynucleotides have one systematic review in aesthetic medicine: nine studies, 219 patients, rated low to moderate quality, with promising results on wrinkle depth, texture and elasticity and no consensus on optimal use^[4]. That is a genuine tier above uncontrolled reports, and well below the PRP literature.

Aesthetic exosome work sits lower again. The 2025 comprehensive review of the field describes early evidence supporting efficacy for alopecia, facial rejuvenation, hyperpigmentation and scarring, drawn substantially from case reports and small studies, and identifies the lack of standardisation in production and application as limiting current use^[5]. A parallel review is more direct: no FDA-approved exosome products exist, isolation methods are inconsistent, source material varies, and clinical trials establishing long-term safety and efficacy have yet to be done^[6]. The measurement problem underneath all of it is why the field publishes and revises minimum reporting standards^[8].

Live-cell aesthetic applications are the hardest to summarise because the clinical literature largely concerns cells as an adjunct to fat grafting rather than as a standalone treatment, and even there the systematic review reports retention rates spanning 25 to 80 per cent and unresolved manufacturing questions^[7].

Three phrases that should slow you down

“Clinically proven.” Not a defined term. Ask which trial, in how many people, against what comparator, measuring what.

“Studies show it stimulates collagen.” Usually true and usually about cells in a dish. Whether it visibly changes a face is a different study.

“Patients report high satisfaction.” Satisfaction is worth measuring and is not an outcome measure of biological effect. Uncontrolled satisfaction data cannot separate treatment from expectation.

Written by the Stem Cell Seoul editorial team and medically reviewed by a Korean Board-Certified Dermatologist (AAD International Fellow · ASLMS member). Last reviewed 2026-08-30.