Technical Note

Plant sterols: one of the best-substantiated claims in the category, and the population where it just failed

Plant sterols carry an authorised cholesterol-lowering health claim built on more than 80 randomised controlled trials, with a dose-response curve and a known plateau. A crossover trial published this month gave 2 g/day for 8 weeks to adults with obesity and found no effect on LDL-C at all (p = 0.632). Both of those statements are true, and the dose was squarely inside the authorised range. This note sets out what is established, what never was, and why “for whom” may be the variable that matters most for this ingredient.

Published 24 October 2026 · 10 min read

What is actually established — and it is a lot

It is worth being precise about how strong this evidence base is, because the rest of the note qualifies it rather than dismisses it.

Dose and effect size. EFSA's panel concluded that at an intake of 1.5–2.4 g/day an average LDL-C reduction of 7–10.5% can be expected, drawing on more than 80 randomised controlled trials. A later assessment of newer data put 1.5–3 g/day at around 11.3–11.4%.
There is a plateau. The effect is dose-dependent but levels off above roughly 2 g/day. This is unusual and useful: the ceiling is known, so “more is better” is answerable, and the answer is no.
Onset and durability. The effect is usually established within the first 2–3 weeks, and trials running up to 85 weeks show it is sustained.
Mechanism is understood. Plant sterols compete with cholesterol for incorporation into intestinal mixed micelles, reducing the fraction of cholesterol absorbed. The site of action is the gut lumen, not the liver — which becomes the whole point later in this note.
The claim is tied to food matrices. The authorised wording was adopted for sterols added to foods such as yellow fat spreads, dairy products, mayonnaise and salad dressings, and intake is with meals.

By the standards of this industry that is close to a best case: a quantified effect size, a defined dose range, a known plateau, a characterised time course, a mechanism, and a regulator-approved form of words. Very few functional ingredients have any of those, let alone all six.

What was never established, even in the general population

Three limits belong on the same page as the claim, because each is routinely blurred in marketing copy:

The endpoint is cholesterol, not cardiovascular events. The claim is about lowering LDL-C — a surrogate marker. There is no outcome trial showing that plant sterols reduce heart attacks, strokes or mortality. Lowering LDL by other means does reduce events, but that inference is borrowed, not demonstrated for this ingredient.
Carotenoid absorption falls modestly. Reduced β-carotene and other fat-soluble micronutrient absorption is a consistent finding, which is why advice to maintain fruit and vegetable intake usually accompanies these products.
Plasma phytosterol concentrations rise. Whether that is itself atherogenic has been argued for years and is not settled. It is a reason for caution in interpretation, not a demonstrated harm.

The new trial: right dose, long enough, no effect

Design

Randomised, double-blind, placebo-controlled crossover trial in 23 adults with overweight or class I–III obesity (BMI 36.4 ± 4.2) and mild hypercholesterolaemia. 2 g/day of plant sterols or placebo, 8 weeks per phase, 6-week washout. Primary outcome LDL-C. Registered as NCT06481020.

Result

LDL-C did not differ between sterol and placebo (p = 0.632). Change from baseline was +0.1 mg/dL (95% CI −7.1 to 7.3) on sterols versus +3.9 mg/dL (−7.6 to 15.4) on placebo. No difference was found in any other lipid parameter, in hs-CRP, or in glucose regulation; no between-treatment difference in body composition, sarcopenia or lifestyle parameters survived correction for multiple testing.

The authors' own reading: this may reflect reduced plant sterol efficacy in this population.

Note what cannot be used to explain this away:

It was not an underdose. 2 g/day sits inside the authorised 1.5–2.4 g range and at the plateau of the dose-response curve.
It was not too short. Eight weeks is three to four times the 2–3 weeks in which the effect is normally established.
It was not a weak design. Randomised, double-blind, placebo-controlled and crossover — each participant serves as their own control, which is the stronger design for a within-person biomarker.

And the honest limits of the trial itself: 23 participants; powered for LDL-C only, so every secondary finding is exploratory; a single centre; and one null result does not overturn 80-plus trials. It bounds them.

The leading explanation — and the study nobody has run

There is a coherent mechanistic reason to expect exactly this result, and it follows from where plant sterols act.

The body gets cholesterol two ways: absorbing it from the gut and synthesising it in the liver. These are reciprocal, and people sit at different points on that balance — a framework often summarised as “absorbers” versus “synthesisers”. It can be measured with non-cholesterol sterol markers: lathosterol tracks synthesis; campesterol and sitosterol track absorption.

The logic is straightforward: plant sterols work by blocking absorption. In someone who absorbs little and synthesises a lot, there is less to block — so the same dose should do less. Reviews using these markers do classify metabolic disorders along an absorber/synthesiser axis, and obesity sits toward the synthesiser end.

But this is a hypothesis, and we should say so clearly: the specific study has not been done. We could not find a trial testing whether the lathosterol/campesterol phenotype predicts plant sterol LDL response in adults with obesity. In men with metabolic syndrome, plant sterols raised plasma campesterol while lathosterol did not change — and those authors themselves called for studies that stratify participants into low and high absorbers. The field has flagged this as an open question, not answered it.

A related caution: commercial “sterol balance” or absorber/synthesiser test panels are marketed with specific percentage-response claims that are not traceable to primary trials. The framework is real; the consumer-facing precision built on top of it is not.

So the defensible statement is: a plausible, testable mechanism predicts reduced efficacy in a high-synthesis phenotype, obesity is associated with that phenotype, and a trial in adults with obesity found no effect. Three consistent pieces. Not yet a demonstrated chain.

What this means for formulators and brand owners

1. The population is the specification, not a footnote

The claim's evidence base is general and mildly hypercholesterolaemic adults. If the target consumer is a weight-management buyer, that is precisely the population where a null trial now exists. An ingredient with excellent evidence can still be the wrong ingredient for a specific product, and that is a formulation decision rather than a scientific dispute.

2. Do not solve a null result by raising the dose

Because the dose-response curve plateaus above about 2 g/day, the usual reflex — it did not work, so add more — has a documented answer here, and the answer is that it will not help. Few ingredients give you that certainty. Use it.

3. Matrix and timing are part of the evidence, not delivery details

The authorised claim was adopted for sterols added to fat-containing foods, taken with meals — which follows from a mechanism that operates on micelle formation during digestion. A capsule taken between meals is not automatically equivalent to a spread eaten with food. Changing the matrix changes the conditions the evidence was generated under, and that has to be its own data. This is the same discipline we have described for route and site of action.

4. Keep the claim on the endpoint that was measured

The authorised claim concerns blood cholesterol. It is not a cardiovascular risk-reduction claim, and no outcome trial supports one for this ingredient. The gap between “lowers LDL” and “protects your heart” is where this category’s copy most often crosses a line it did not need to cross — the cholesterol claim is strong enough on its own.

5. Treat “who responds” as the live research question

If absorber/synthesiser phenotype does turn out to predict response, that is a responder-stratification story rather than a reason to abandon the ingredient — and the stratified trial is still available to whoever runs it. Until then, do not build consumer-facing personalisation on a framework whose predictive step is untested.

A note on what we do and do not supply

We do not supply plant sterols. Our catalogue is microbial — lactic acid bacteria, bifidobacteria, Bacillus, yeasts and custom blends. This note exists because the ingredient is an unusually clean case study in how an evidence base can be strong and still have a population boundary, which is a question we get asked about our own category constantly.

If you are assessing a microbial ingredient and want the same treatment — what the evidence establishes, in whom, by what route, at what dose, and where it stops — that is a conversation we are glad to have. We will not use any study to support an efficacy claim for a product, ours or anyone else’s.

References

Dietary supplementation with plant sterols for obesity management and related metabolic disorders: A randomized, double-blind, placebo-controlled crossover clinical trial. The American Journal of Clinical Nutrition, 2026. ClinicalTrials.gov NCT06481020. PubMed record
EFSA Panel on Dietetic Products, Nutrition and Allergies. Scientific Opinion on plant sterols/stanols and blood cholesterol — the basis of the authorised claim, concluding an average 7–10.5% LDL-C reduction at 1.5–2.4 g/day from more than 80 randomised controlled trials. EFSA statement
Commission Regulation (EU) No 376/2010, adopting the authorised wording and the food categories to which the claim applies. Text
Non-cholesterol sterol concentrations as biomarkers for cholesterol absorption and synthesis — the lathosterol / campesterol framework and its use to classify metabolic phenotypes. Record

This note summarises and interprets published research and regulatory opinion in order to explain how population, dose, matrix and endpoint bound the applicability of evidence. It is not a claim about the efficacy or safety of any product, is not directed at any company or brand, and is not medical, dietary or regulatory advice. Baifude does not supply plant sterols. Authorised claim wording and permitted food categories differ between jurisdictions; the current regulation in the market concerned governs in every case. Findings apply only to the populations, doses, matrices and conditions described in the original sources.