Centella asiatica

The proposed mechanisms are wide-ranging. It is worth separating what has been shown in people from what has been shown in a dish.

Collagen. The best-known claim, and the shakiest. Skin cells in a dish do make more collagen when centella is added. But the two main studies disagree on which compound does it. One found asiatic acid was the only active part. The other found asiaticoside and madecassoside both worked. That contradiction is almost never mentioned. There is also a larger body of scarring research where these same compounds reduce collagen. Taken together, that looks more like modulating tissue repair than simply building more. No human study has shown a collagen increase on biopsy.

Inflammation. Real and reproducible — in mice and rats, at oral doses. No human study isolates an anti-inflammatory effect of centella from the other ingredients it was tested alongside.

Barrier. The claim that centella raises filaggrin, aquaporin or tight-junction proteins has no primary source that could be found. The one genuine mechanistic study fed a standardised extract to mice for eight weeks and found improved hydration and restored ceramide production. That is oral, in animals, with the whole extract.

A problem nobody markets around. Madecassoside and asiaticoside are big, water-loving molecules. At around 960 to 975 daltons, they sit well above the rough 500-dalton ceiling for crossing the outer skin layer. The drug-delivery literature says so plainly and often. A whole research field exists just to encapsulate these compounds, because they do not cross skin well on their own. So the two compounds cosmetics promote hardest are the two least likely to reach the cells the lab studies used.

The names on labels do not all mean the same thing.

• Centella asiatica extract — the generic cosmetic ingredient. Unstandardised, mostly solvent, batch-variable.
• Madecassoside — a named single compound. If a product states a percentage of madecassoside, you at least know a dose. It is the most abundant triterpene in the plant and the one excluded from the pharmaceutical extract.
• Asiaticoside — another named single compound, and the one with the cleanest single human trial behind it.
• Asiatic acid and madecassic acid — the smaller, sugar-free versions. Roughly half the size, and therefore the most likely of the four to actually penetrate.
• TECA / TTFCA — titrated extract of Centella asiatica. The defined pharmaceutical mixture used in licensed wound ointments.
• "Cica" products — a marketing category, not an ingredient. Some contain very little centella, and some barrier creams sold as cica contain none at all, relying on entirely different actives.

The practical rule: unless a product names a specific compound at a stated percentage, or names a standardised extract, the number on the front does not tell you the dose of anything active.

There is no meaningful standard concentration, and that is the honest answer.

• 1% TECA is the strength in licensed medicinal ointments in Europe. That is 1% of a defined triterpene mixture.
• 2.5% and 5% extract were the concentrations in the cleanest human barrier study, which found 5% gave the better result on hydration and water loss.
• 0.1% madecassoside was used in the most-cited photoaging study — alongside 5% vitamin C, with no madecassoside-only comparison, so nothing there can be attributed to the centella.
• 0.05% standardised extract was used in the best-designed post-laser trial.

Cosmetic products advertising "5% cica" or "10% Centella" are quoting a percentage of an extract that is mostly solvent. It is not comparable to any of the figures above.

Centella is a slow, quiet ingredient. Expect a gradual reduction in tightness and a calmer feel over two to four weeks, in the way a good moisturiser works. Nothing dramatic.

For post-procedure use, the one properly controlled trial found redness measurably lower on the treated side after laser resurfacing at days 2, 4 and 7. The effect was small — the difference on the redness instrument barely cleared statistical significance, and other skin measurements showed no difference at all.

On tolerability, the honest picture is more complicated than the reputation.

At cosmetic strengths it is very well tolerated. Patch testing across several hundred people found no meaningful irritation or allergy at 0.045% to 0.5%. In animal testing it was classed as a very weak sensitiser.

But centella is a confirmed contact allergen, with case reports published from 1985 onwards. About ten cases exist in the world literature. Two things about them are worth knowing:

• Almost every case involved medicine-strength preparations on damaged skin — surgical wounds, keloids, burns, graft sites, leg ulcers. That matters, because applying cica products to raw or post-procedure skin is exactly what people now do.
• Fragrance is a frequent co-factor. Lavender and geranium oils appear repeatedly in these case reports. A reaction to a cica product is at least as likely to be the fragrance as the centella.

Nobody knows how common centella allergy is. No one has ever patch tested a run of consecutive patients for it, so any percentage you see quoted has no source behind it. There is also no ready-made patch test for centella, which makes it harder to diagnose than most.