The present dossier for glyphosate is comprehensive. However, it does not include a DNT study that is compliant with test guidelines.
A DNT study is however available for the active ingredient glyphosate trimesium . It was performed in 2001, and sponsored by the company that had submitted the dossier for the trimesium salt for the first EU-wide evaluation. We learned about its existence in March 2022, and immediately informed EFSA. EFSA confirmed that the DNT study was neither included in the present nor in previous dossiers.
The DNT study has been evaluated by the U.S. EPA in 2005, and the US authority concluded that it demonstrates behavioural effects in rat offspring following exposure to maternal animals . The doses were 0, 10, 25 and 100 mg glyphosate trimesium/kg body weight (bw)/day, administered to maternal animals from gestational day 7 through postnatal day (PND) 11 by gavage. The maternal lowest observed adverse effect level (LOAEL) was > 100 mg, i.e. no maternal toxicity, deemed as adverse, was noted. In the offspring, overall motor activity was decreased (45-72%) in males and females in the 25 and 100 mg groups on PND 14. These results were already recognised in the original study report from 2001 but dismissed by the test laboratory as incidental. In contrast, the U.S. EPA acknowledged these effects and set a LOAEL in offspring at 25 mg/kg bw/day and the no observed adverse effect level (NOAEL) at 10 mg/kg bw/day. So, we note that the interpretation of the data by the test laboratory differs from the U.S. authority’s interpretation; effects recognised but dismissed by the laboratory were used by the U.S. EPA to set a LOAEL. Overall, the U.S. EPA also deemed the study acceptable for regulatory use .
At that time, the acceptable daily intake in the EU was based on a NOAEL of 31 and 21 mg/kg bw/day for glyphosate and glyphosate trimesium, respectively; both were based on chronic toxicity studies in rats.
As citizens and scientists, we would expect that if one glyphosate salt is found to cause DNT at a dose level thought to be safe for other glyphosate salts, then it would have to be clarified, without unnecessary delay, if other variants of glyphosate share that property. This is because it could be the glyphosate molecule itself causing the effect, and use as well as human exposure to glyphosate is widespread . Indeed, several provisions in EU’s pesticide regulation would serve this purpose. We see at least three violations of these provisions:
1. The company should have submitted the DNT study directly in 2001
Pesticide legislation in force in the EU in 2001  stipulated that the holder of an authorization must “immediately notify the competent authority of all new information on the potentially dangerous effects of any plant protection product […]". Compared to other effects of glyphosate trimesium relied on in the original evaluation for setting the acceptable daily intake (ADI) , the effect in the DNT study was observed at a lower dose level. Therefore, a consideration of the DNT effect could have reduced the ADI already at that time.
Importantly, the DNT effect was recognised by the test laboratory in the original study report, as cited in the evaluation by U.S. EPA . The observed effects were interpreted as incidental by the test laboratory and thus dismissed; nonetheless, we claim that the results as such still indicate potentially dangerous effects. Companies may, and often do, argue that certain observed effects are not relevant or reliable. Any final decision on dismissing apparent effects as incidental must however be made by authorities.
Considering that the effect dismissed by the test laboratory in this case was dose-dependent, consistent between sexes, and substantial in magnitude, a bias in the interpretation by the test laboratory cannot be ruled out. Needless to say, a bias in data interpretation that results in an underreporting of adverse effects will negatively affect the authorities’ ability to protect public health . This is however not the main subject of the present paper.
As far as we could establish, authorities of the EU or its member states were never informed of the existence of the effects observed in the industry-sponsored DNT study of glyphosate trimesium , although products containing this substance were authorised at that time.
It seems therefore as if the requirement to notify the competent authorities on the observed DNT effects in 2001 has not been fulfilled.
2. The company should have submitted the DNT study for the ongoing re-evaluation
Glyphosate trimesium is highly water-soluble and dissociates fully in water  and thus also in the body. Conceptually, the observed DNT effects could then have been caused by the glyphosate molecule, or by the trimesium ion, or possibly by both in combination.
In the EU, an active substance shall be approved “if it may be expected, in the light of current scientific and technical knowledge”, that its proper use does not cause harmful effects on human health. It is the responsibility of the applicant to demonstrate this in the dossier. For the present case, at least one of the applicant companies had scientific knowledge that the glyphosate molecule, i.e. the active substance in the present dossier, was among very few potential causes of DNT effects in the study of glyphosate trimesium. It is therefore counter to the intentions of the law and the responsibilities of the applicants to assume that the glyphosate molecule has not caused the observed DNT effects. To make that assumption it must be established that the trimesium ion, or trimesium and glyphosate in combination, were the causes, or the glyphosate molecule has to be cleared by other evidence.
It is the responsibility of the applicant companies to appropriately make use of this scientific knowledge . It can be no-one else’s responsibility because no-one else involved in the regulatory process had access to this knowledge.
In principle, the applicants could consider the glyphosate trimesium DNT study in three ways for the present assessment of glyphosate:
First, the study could be used as such to assess and characterise DNT of glyphosate and the salts currently under assessment, in conjunction with academic animal and epidemiological studies of DNT-related effects from glyphosate or its formulations .
Second, it could trigger the conduct of a new DNT study for glyphosate or one of the salts currently under evaluation (see also next section).
Third, it could be disregarded because the DNT observations were considered irrelevant, e.g. since they can be attributed to the trimesium ion and hence not expected to manifest with other salts. Or because the company, for some reason and in contrast to the U.S. EPA, finds the study unreliable.
It is however not clear if the applicants have considered or acted upon any of these options, as there is no reference to this study in the present glyphosate dossier.
The legislation requires companies to submit sufficient information for the evaluation of foreseeable risks, as well as “[a]ny information on potentially harmful effects of the active substance” on human health to EFSA . We argue that, in whatever way the applicants decide to make use of the DNT study of glyphosate trimesium in their assessment of glyphosate, the regulatory authorities must be in a position to review that assessment and to make the final decision on how to use – or not – the existing DNT study in the present assessment of glyphosate. Therefore, the companies’ obligation to transparently report their use of the DNT study and to make the study available to EFSA is evident.
Certain differences in the toxicological profile of glyphosate and glyphosate trimesium have been recognised by EU authorities during the first evaluation over 20 years ago [5, 13]. In particular, the acute toxicity of glyphosate trimesium was substantially higher; the long-term toxicity was similar between the different forms. At that time, it was concluded that data for glyphosate trimesium should not be used for evaluation of glyphosate. This reasoning was however based on a situation where studies for each type of toxicity were available for both forms of glyphosate. The discussion did not include any reasoning or guidance on how to proceed if data indicate an adverse effect of one form of glyphosate on an endpoint where data were lacking for the other form. Pesticide regulation including the data requirements have also changed substantially since then. In any case, it would be up to EFSA, and not a company, to decide if such earlier reasoning would be applicable today in the present case.
So, it seems that this company-owned DNT study  should have been considered in the ongoing renewal process, and we therefore find it improper that it was omitted from the dossier submitted to EFSA.
3. The present dossier should have addressed DNT
For the evaluation of reproductive and developmental toxicity, there is a requirement that companies’ “[i]nvestigations shall take account of all available and relevant data, including […] knowledge concerning structural analogues to the active substance”. Also, “[p]otential neurotoxic […] effects […] shall be carefully addressed and reported” in these investigations .
Accordingly, in case the company sees no direct applicability of the glyphosate trimesium DNT study for other forms of glyphosate, they would still have been under obligation to consider that study and to “carefully address” DNT of those other forms of glyphosate. It is of course a matter of interpretation what it means to “carefully address” an endpoint. But one feasible option could be to commission a new DNT study. A discussion or conclusion regarding DNT of glyphosate was absent from the present dossier. Also, results from literature searches regarding effects of glyphosate on autism or ADHD risk were not included in the dossier, as already highlighted by AGG .
We note that the EU assessment report negated the need to perform a DNT study for glyphosate . This conclusion was however not informed by results from the DNT study of glyphosate trimesium. According to the data requirements, “when indicated by observations in other studies or the mode of action of the test substance, supplementary studies or information may be required to provide information on the postnatal manifestation of effects such as developmental neurotoxicity.” .