Below the TDI, Above the Limit: What EFSA Actually Decided on S-Acetyl Glutathione

EFSA calls the novel food safe. It also cuts the applicant's target population, rewrites its specifications, and clears a contaminant carried at roughly eighty times any permitted food concentration.

On 2 July 2026 the EFSA Panel on Nutrition, Novel Foods and Food Allergens adopted its opinion on S-acetyl glutathione, an acetylated tripeptide submitted by Gnosis S.p.A. under Article 10 of Regulation (EU) 2015/2283. The conclusion is a clearance. The novel food is safe at 300 mg per day for adults, excluding pregnant and lactating women.

That sentence is the part that will be quoted, and it is the least informative part of the document. Read the reasoning and three separate corrections to the dossier emerge, each of which will shape the eventual Union list entry more than the verdict does.

A contaminant cleared eighty times above the food limit

The novel food carries perchlorate. Across 25 batches it averaged 26.7 ± 7.20 mg/kg, and the specification limit is set at 60 mg/kg. For comparison, the highest maximum level for perchlorate anywhere in Regulation (EU) 2023/915 is 0.75 mg/kg, for dried tea leaves.

EFSA did not apply that maximum level. It calculated intake at the specification limit and compared the result with the tolerable daily intake of 1.4 µg/kg bw per day set by the CONTAM Panel in 2025. For adults, the novel food contributes 0.26 µg/kg bw per day against a background of 0.21, well inside the TDI. For young adolescents the figures are 0.41 and 0.50, a combined 0.91, or roughly 65% of the TDI before anything else in the diet is counted.

The two regimes are measuring different things. Contaminants law regulates concentration in a food because it cannot know how much of that food anyone eats. Novel food law regulates a defined daily dose, so it can compute the mass directly. When the vehicle is 300 mg of powder, an extreme concentration yields a trivial mass. Both answers are correct within their own logic. They are simply not commensurable.

The consequence is practical rather than theoretical. Anyone reading the Union list entry will see a novel food authorised with a perchlorate specification two orders of magnitude above a concentration that would be unlawful in tea. Explaining that to a national control authority, to a customer's quality department, or to a journalist, is a problem the authorisation does not solve. And the headroom is thinnest precisely where the evidence is weakest.

The applicant asked for one population and received three

Gnosis proposed 300 mg per day for the general population above 10 years of age. The Panel did not refuse. It re-derived the number. From a BMDL10 of 877 mg/kg bw per day and an uncertainty factor of 200, it arrived at a safe dose of 4.39 mg/kg bw per day, which converts by body weight into 190 mg per day for young adolescents aged 10 to under 14, around 269 to 270 mg per day for older adolescents aged 14 to under 18, and 307 mg per day for adults.

So the applicant receives its requested figure only for adults. The opinion itself carries the rounding forward inconsistently, reporting 269 in the abstract and 270 in the conclusions. That is a small thing, but it is the kind of small thing that must be resolved before it becomes a figure in an implementing regulation.

Two consequences follow. The conditions of use will need age-differentiated maxima and corresponding labelling, which is a formulation and packaging question, not a legal footnote. And the reason for the split is not a toxicological signal in adolescents. It is the absence of one. There are no human data between 10 and 18 years. The Panel accepted ten human intervention studies of oral glutathione in adults as supporting evidence, then extrapolated to the younger bands on a body weight basis, assuming comparable tolerance.

Where the uncertainty factor came from, and what it cost

The factor of 200 decomposes into ten for interspecies variability, ten for intraspecies variability, and two for extrapolating from a subchronic to a chronic exposure. That last factor exists only because the pivotal study is a 90-day rat study. It is also the factor that halves the safe dose, and therefore the factor that produced the adolescent carve-out. Without it, the safe dose would be 8.78 mg/kg bw per day, which delivers the requested 300 mg comfortably across all three age bands.

That is a dossier design lesson with a value attached. The toxicology package here was well built and raised no genotoxicity concern. It was one study duration short of the evidence that would have avoided a commercially awkward restriction.

Adaptive or adverse: the line was drawn twice

The 90-day study produced organ weight changes in two organs. In the kidney, relative weight rose dose-dependently, urine volume increased by up to 151% in males, and urinary pH fell. No treatment-related macroscopic or histopathological change was seen at necropsy. The Panel called this adaptive.

In the liver, absolute and relative weight rose dose-dependently. No clinical pathology parameter was altered, and no macroscopic or microscopic liver change was reported at necropsy at any dose. The Panel called this adverse, and made it the reference point for the entire assessment.

The distinguishing criteria were explicit: magnitude above 10% at the highest dose, persistence into the recovery group for the liver-to-body weight ratio in males, and values falling outside the laboratory's historical control range. Not pathology.

This deserves attention beyond one supplement ingredient. Organ weight change without a morphological correlate is among the most contested categories in regulatory toxicology. Here the Panel resolved it with a reproducible three-part test, anchored the benchmark dose analysis on it, and cited the public repository of BMD analyses in support. Whether or not the liver call persuades you, reasoning of that shape becomes precedent for the next dossier.

The premise that is not in the file

S-acetyl glutathione is sold on the proposition that acetylation improves delivery of glutathione. The safety dossier contains what is now the only publicly available, regulator-reviewed comparison. In a crossover study at 3.5 g, some 11.5 times the proposed daily dose, S-acetyl glutathione was not detected in any plasma sample. Plasma glutathione Cmax and AUC were higher after the acetylated form; glutathione in red blood cells was higher after plain glutathione. The Panel's own summary is that the glutathione concentration-time profile derived from the novel food did not substantially differ from that of glutathione given directly.

Two caveats are essential. EFSA states expressly that it is assessing risk, not efficacy or any claimed benefit. And the study did not compare equimolar amounts, with the glutathione dose delivered via the acetylated form running about 14% lower. Nothing in the opinion disproves the product's premise.

It does, however, illustrate something structural. A novel food authorisation is a safety instrument, yet safety files are increasingly the richest public evidence base on ingredients whose commercial propositions are never separately adjudicated, because no health claim was ever sought. The file that clears a product for sale is also the file a competitor, a control authority, or a claims regulator will open when the marketing runs ahead of the data.

What the routine files decide

The Panel also cut the arsenic specification from the proposed 1.5 mg/kg to 0.5 mg/kg on the strength of the applicant's own analytical data, and corrected an assay ceiling that had been proposed above 100%. Every element of the dossier was found pertinent to the conclusion under Article 26, so the applicant should expect data protection to attach to the authorisation.

None of this is procedurally novel. The methodology is the standard NDA guidance. Glutathione has been on the EU supplement market for years, has been generally recognised as safe in the United States since 2009, and was assessed by ANSES in 2008. The novelty is confined to the acetylated molecular structure and the synthesis route, under points (i) and (vii) of Article 3.

Which is rather the point. Routine opinions are where the operative decisions live. This one will determine what a company may put on a label for three different age bands, what concentration of a thyroid-active contaminant is lawful in a supplement but not in tea, and what counts as an adverse effect when the microscope shows nothing at all. The verdict took one sentence. The decisions took seventeen pages.

Source: EFSA NDA Panel, Safety of S-acetyl glutathione as an novel food pursuant to Regulation (EU) 2015/2283, EFSA Journal 2026;24(9):e10258, adopted 2 July 2026. DOI: 10.2903/j.efsa.2026.10258. Question number EFSA-Q-2024-00633.