The fate of a botanical extract under Annex III is not decided in the laboratory, by its toxicological profile, but by the procedural channel it is in. That channel determines something more important than the prohibition itself: whether the operator keeps the right to have its evidence examined.
Substance or preparation
In our analysis of hydroxyanthracene derivatives we argued that the conflict is not merely scientific but one of legal basis. Article 8 of Regulation (EC) No 1925/2006 allows the inclusion in Annex III of a substance — an identified chemical compound — or of an ingredient containing it. Regulation (EU) 2021/468 went further: in addition to prohibiting specific substances — aloe-emodin, emodin and danthron — it prohibited a whole series of botanical preparations containing them, namely Aloe leaf extracts and preparations of Rheum root, Cassia senna leaf, and Rhamnus frangula and Rhamnus purshiana bark. Those preparations are not “substances” or pure chemical compounds in themselves, but mixtures of variable composition obtained from a plant through an extraction process, in which the hydroxyanthracene derivative coexists with many other compounds, in proportions ranging from significant amounts to mere traces. Prohibiting aloe-emodin as a chemical substance and prohibiting every aloe leaf preparation containing it are therefore two different measures.
In November 2024 the General Court held that Article 8 authorises only the first. It expressly confirmed that aloe-emodin and emodin are “substances” for the purposes of Article 8 (T-302/21, paragraph 40), so their prohibition was not annulled. But preparations — those mixtures of variable composition — fit neither the concept of “substance” nor that of “ingredient containing a substance”. The Court reasoned that, by using those two words, the legislature chose concepts of limited scope. Had it wished to allow entire preparations to be prohibited, it would have said “preparation”. As it did not, the Commission exceeded its powers in prohibiting not only the substance but every preparation containing it. Hence the annulment of the first, second and third entries of Article 1(1) and of Article 1(2) of the contested regulation.
The thesis we defend in this article is not a scientific novelty but a procedural trap: the fate of an extract under Annex III is not decided in the laboratory — it does not depend on its toxicological profile — but by the procedural channel it is in. From that administrative position follows something far more significant than the prohibition itself: whether or not the operator keeps the right to have its evidence examined. Article 8 recognises that right at a single moment, the Union scrutiny phase, and for a very short period. Before that phase there is no procedure to turn to, because no file is open; afterwards there is none either, because the route is closed. The consequence for operators is clear. The decisive moment in the whole process is not when the extract is prohibited, but a much earlier and far less visible one: when the period for defending it opens.
That is why an operator's first decision should be diagnostic rather than strategic.
Diagnosis: where the extract stands in Annex III
The diagnosis consists of one thing only: knowing in which part of Annex III the extract is. Being in Part A (prohibited), Part B (permitted with restrictions) or Part C (under Union scrutiny) is not the same. In each of these three situations the operator has a different margin for action, and the difference is radical.
If it is listed in Part A of Annex III, its addition to foods and its use in their manufacture are prohibited, and the operator has no formal channel to reverse that. The reason is that Article 8(4) recognises the right to submit a safety file only for substances listed in Part C. For Part A there is no equivalent provision and, consequently, no right of submission, no deadline for the Authority to give an opinion, and no decision the Commission is obliged to adopt. That does not make the entry unchangeable: Annex III is amended under that same Article 8, and nothing prevents an entry from being reviewed if new data emerge. But the operator can only send information, ask its national authority to raise the matter in committee, or promote a review through its trade association. The difference is the one between a right of initiative and a power of persuasion.
If the extract is listed in Part C, it may continue to be marketed and the operator keeps that right of initiative, but subject to a strict time limit. Article 5(2) of Implementing Regulation (EU) No 307/2012 only admits files submitted within twenty-four months of the entry into force of the listing, and Article 8(5) of Regulation (EC) No 1925/2006 requires a decision within four years of that same listing. Add the nine months the Authority has to deliver its opinion under Article 6(2) of Implementing Regulation (EU) No 307/2012, as amended by Implementing Regulation (EU) 2023/612, and the opinion lands close to the four-year limit, leaving a timetable with no slack. Those twenty-four months are therefore, for any extract under scrutiny, the most important date in the whole file. Once they have passed without evidence being submitted, the operator may keep the product on the market for the time being, but will have lost the only channel in which its defence had an addressee obliged to examine it.
Finally, if the extract appears in neither list — which is the position of the vast majority of botanical extracts — there is no European restriction to apply, but no authorisation to invoke either. Its use is governed by national rules, with the market fragmentation that entails.
This third position deserves attention, because it is the most common and the least understood. Absence from Annex III is usually read as legal certainty, when it actually means something different: not a presumption of safety, but the absence of any pronouncement by a European authority. And the hydroxyanthracene derivatives file shows how easily that limbo can end. It was enough for some Member States to express concerns about a potential risk during the assessment of a health claim. That mere concern triggered the scrutiny procedure of Article 8(2), whose threshold does not require certainty but is satisfied by the mere possibility of harmful effects and the existence of scientific uncertainty. No new toxicological finding or market alert was needed — only a discussion that had begun for a reason unrelated to the safety of these derivatives.
One consequence must be stressed: for an extract to come under scrutiny, it is enough for any procedure to bring it before the Authority or the committee.
That exposure can come about in two different ways. The first is from the sector itself, as happened with hydroxyanthracene derivatives: a health claim application submitted under Article 13(5) of Regulation (EC) No 1924/2006 ended up triggering the Article 8 procedure. The second is institutional: Article 8(2) allows any Member State to ask the Commission to start the process. This was the case with Regulation (EU) 2015/403, which placed Ephedra herb in Part A and Yohimbe bark and its preparations in Part C, after an opinion concluded that the available chemical and toxicological data were not sufficient to ensure their safety.
The point is not to discourage procedures that companies need or are required to undertake. It is to warn that all of them open a file before institutions that may ask very different questions from the one originally raised. Anyone about to start such a procedure would do well to anticipate how they would respond if the question suddenly shifted from efficacy to safety.
Composition: the only variable the operator controls
Once the extract has been located, the next key question is its composition. It should be addressed before any other, since it is the only one the operator can resolve on its own and with immediate effect on the market. The Annex III entries on botanicals do not prohibit the plant as such, but the preparations containing the derivative concerned. The nuance is not rhetorical: recital 10 of Regulation (EU) 2021/468 expressly recognises that these derivatives can be removed from preparations by filtration processes, yielding products that contain them only as trace impurities. Inner leaf gel and decolorised extracts that demonstrate that absence therefore fall outside the literal scope of the prohibition. The commercial dividing line is drawn by the production process, not by the plant species.
Sustaining that position, however, is not a matter of legal interpretation but of proof. It requires a validated analytical method, a specification with defined limits for the relevant markers, and documentary traceability capable of demonstrating conformity batch by batch. An operator with that technical support will be disputing an analytical result in an inspection; one without it will be disputing a prohibition, which is the worst procedural position imaginable. The difference between the two scenarios is determined not by the quality of the legal argument but by that of the technical file behind it.
A warning should be added that is rarely voiced. The EFSA opinion of 22 November 2017 observed that hydroxyanthracene-depleted aloe extracts contained an additional genotoxic component. So although the composition boundary is legally sound — it rests on the wording of the Annex III entry, which prohibits preparations containing the derivative, and on a recital of the regulation itself — it is not immune to future assessment. Anyone building a strategy on it should know that its strength is regulatory, not scientific; in other words, the boundary may shift if the Commission decides to reformulate the entry.
Characterisation: an anticipated burden of proof
All of the above leads to a requirement common to the three positions, which should be stated precisely. Outside the novel food regime, no rule imposes an obligation to characterise or standardise a botanical extract. The real requirement comes from another source: the guidance the Authority addresses to industry and the practice of its own opinions. These are the texts that set the level of botanical identification, production process and detailed composition against which a file will be assessed. Admittedly, such guidance is not binding — as the General Court recalled when it denied interpretative value to the 2009 EFSA guidance — but in practice it determines the outcome.
Yohimbe is the best example. It entered the scrutiny procedure because its chemical and toxicological characterisation did not allow a view to be taken on its safety, and from there it moved straight to prohibition without any new negative data emerging. Conversely, the studies the sector submitted to defend the other species were discarded, among other reasons, because they had been carried out on preparations with a low content of the derivative under examination.
This is therefore not a legal obligation but an anticipated burden of proof. An operator that decides not to characterise its extract today is not breaking any law; it is simply putting itself in a position where it cannot demonstrate safety on the day it is asked to. And when that day comes, twenty- four months will be wholly insufficient to generate from scratch the scientific data then required.
With the position diagnosed and the composition documented, what remains is the question that consumes most resources and is decided worst: which route to take, which sales channel to use and where not to invest. That is the subject of the second part.
Javier Gómez Sanz is the founder and director of Food Tech Legal. Comments on this analysis may be sent to editorial@foodtechlegal.com. Enquiries about a specific product or case are handled through the quote form. The original Spanish version of this analysis is available here.