Our Best-Scoring Cream Has Five Ceramides and No Cholesterol

We sell a cream with five ceramides in it. It scores 9.8, the highest grade we give, and the box says 5-Ceramide Complex in the place where a box says the thing it wants you to remember.

Last week we went through all fifty-four products counting ceramides, because a customer asked whether five was better than one and we realised we had never checked. Eleven products have a ceramide. Five of those eleven carry the same set of five — NP, NS, AS, AP, EOP, always in that group, never a different combination. The other six carry Ceramide NP on its own. Nothing in our catalogue has two ceramides, or three, or seven.

So the answer to "is five better than one" starts by admitting that five and one are the only two options anyone is selling, and that neither number came from the skin.

Eleven products out of fifty-four. Five with the full set, six with Ceramide NP alone. The six singletons are scattered across categories that have nothing to do with each other: a jojoba oil with eleven total ingredients, a cleansing milk, a body cream, a sheet mask, a gel cleanser, and — this one surprised us — a hair pack.

Zero pseudo-ceramides. We looked, because pseudo-ceramides are common in Korean barrier creams and we assumed we stocked some. Hydroxypropyl Bispalmitamide MEA appears nowhere in our fifty-four. It appears in a competitor's list we hold in our own database, which is a detail we will come back to.

Five is a chromatography artifact

Here is where five comes from. The answer is 1991, and it is not flattering.

In 1991 Genji Imokawa's group stripped forearm stratum corneum from 65 healthy people and 32 to 35 people with atopic dermatitis, ran it on thin-layer chromatography, and separated six ceramide fractions. Ceramide 1 was the most depleted in atopic skin, in lesional and non-lesional sites both. It is a good paper and it founded the entire ceramide-moisturiser category.

Six was what TLC could resolve in 1991.

By 2009 Yusuke Masukawa's group was running eleven named classes on LC-MS, monitoring 182 molecular ions at once, drawing on prior structural work that had already catalogued 342 distinct ceramide species. Their forearm measurements show the classes are nowhere near equal: in one sample CER[NP] came in at 11.5 ng per µg of protein and CER[ADS] at 0.24, a roughly fiftyfold spread between the largest and smallest class in the same piece of skin. Two years later Jeroen van Smeden's group wrote that "in human SC, the literature currently reports 11 CER subclasses," then identified a twelfth. In 2012 a Ghent group profiling tape strips by reversed-phase LC-QTOF found classes nobody had described, including one with four hydroxyls on the sphingoid base, and estimated that human stratum corneum contains over 1,000 unique ceramide structures.

Which means a five-ceramide complex covers, generously, five of eleven or twelve subclasses and five of several hundred species. That is not a criticism of the formula. It is a correction to the arithmetic the box invites you to do.

Where the five actually sit

Position on an ingredient list is the only concentration signal a consumer gets, and it is a weak one. Under 21 CFR 701.3, ingredients above 1% must be listed in descending order of predominance. Below 1%, they may be listed "without respect to order of predominance." The label stops carrying information at exactly the concentration where ceramides usually live.

Read the last row again. Four of that product's five ceramides are the last four ingredients on the list. They are the end of the label. Whatever is in there is under 1%, and under 1% the order they printed is legally arbitrary, so we cannot even tell you which of the four there is most of.

The foam cleanser at positions 10 through 14 is doing something different, and it is a rinse-off product, which cuts the other way. We have written before about what label position can and cannot tell you about concentration, so we will not run that argument again here.

The 1996 paper that made us recount

Then we found Man, Feingold and Elias, and the question changed shape.

Mao-Qiang Man, Kenneth Feingold and Peter Elias spent the early nineties applying lipids to damaged skin and measuring how fast the barrier came back. Their 1996 paper in the Journal of Investigative Dermatology reports the result in one sentence: application of one or two of these lipids to perturbed skin delays barrier recovery; only equimolar mixtures allow normal recovery. The three lipids are ceramide, cholesterol and free fatty acid. Give the skin one of them and repair goes slower than giving it nothing. Acylceramides on their own — the EO class, the one in every five-ceramide kit — delayed recovery specifically. Get all three in, and further acceleration follows as you enrich any single component up to threefold.

The 1993 Archives of Dermatology paper that preceded it gives the mechanism: incomplete mixtures produce abnormal lamellar bodies, and those build abnormal intercellular membrane sheets in the stratum corneum. It is not that the incomplete mixture fails to help. It is that it gets incorporated and builds the wrong structure.

So we ran our eleven ceramide products against that criterion. Ceramide, plus cholesterol, plus a free fatty acid, all three present.

Two of eleven pass.

The Nomadia cleansing milk has Ceramide NP at 15, Cholesterol at 23 and Linoleic Acid at 27. The Birthtii body lotion has all five ceramides, Cholesterol at 36, and Palmitic and Stearic Acid at 8 and 10 — although our own data files those two as "Surfactant" and "Emulsifier," and in a soap-based emulsion they may well be saponified rather than sitting free. We don't know. Nobody publishes that.

Our 9.8-scoring CELLEANSER Derma Barrier P Cream has five ceramides, no cholesterol and no free fatty acid. It has Beta-Sitosterol at 23 and Phytosterols at 26 — plant sterols, structurally cholesterol's cousins, and the brand's own copy calls them "cholesterol-like." Whether a phytosterol substitutes for cholesterol in the Man model has not been tested that we can find. Man's group did test what does and doesn't substitute, and found glycosylceramides and sphingomyelin could stand in for ceramide while cholesterol esters could not stand in for cholesterol. Plant sterols were not in that experiment.

What the rivals have

We keep 327 competitor records with our own scores attached, and this is where it gets uncomfortable.

Our benchmark note on the ILLIYOON Ceramide Ato Concentrate Cream, written by us, says it is "built on a pseudo-ceramide where ours uses five real ceramides." The ingredient list we stored alongside that note lists a pseudo-ceramide, and Ceramide NP, and Cholesterol, and Phytosphingosine. On the Man criterion — the one we just spent four paragraphs calling the criterion that matters — ILLIYOON's lipid stack is more complete than ours. We wrote a dismissal of a product that has the ingredient our product is missing.

The scores are their own problem. That same ILLIYOON cream appears four times in our database at 9.5, 9.7, 9.1 and 8.8, and three of those four came from the same source. A 0.9-point spread on one product. Aestura Atobarrier365 Cream sits at 9.7 and 9.2. Round Lab 1025 Dokdo Cream at 9.5 and 9.1. Dermalogica Daily Microfoliant at 8.4 and 9.1, both from Amazon. We published the full audit of how our scoring goes wrong and this is the same disease showing up in the ceramide aisle.

The highest-scoring ceramide rival we hold is Vanicream Daily Facial Moisturizer at 9.7, filed under our own label "five-ceramide fragrance-free barrier cream." Across all 327 records the two platforms we pull from average 8.68 and 8.71. That difference is 0.03 and it means nothing.

What our screening found

Standing in front of the shelf

Stop counting ceramides. The count tells you which supplier kit the formulator bought.

Look for cholesterol in the list instead, and for a free fatty acid — linoleic, palmitic, stearic. If a product has ceramides and neither of the other two, the strongest evidence we have says the mixture is incomplete, and incomplete mixtures did worse than nothing in the animal work. If all three are there, you have a formula built by someone who read the barrier literature rather than the catalogue.

And check where they sit. Ceramides at position 40-something of 46 are a claim. Ceramides in the top fifteen are a formula. Our own screening and award records are public for exactly this kind of second-guessing, and this piece is us doing it to ourselves.