TFA and EU Food Safety Governance: Risk, Hazard and MRLs—All at Once

Within seven weeks, TFA has been hit from two different regulatory directions -one measuring dose, the other declaring hazard. Where they meet is where MRL policy gets decided.

On 20 July, EFSA published its updated health-based guidance values (HBGVs) for trifluoroacetic acid. This came just over six weeks after ECHA’s Risk Assessment Committee recommended, on 5 June, classifying TFA as toxic to reproduction, Category 1B (H360Df), for developmental effects -alongside PMT and vPvM.

EFSA lowered the acceptable daily intake from 0.05 to 0.014 mg/kg of body weight per day and established an acute reference dose of 0.07 mg/kg for the first time. These more stringent benchmarks can inform EU decisions on dietary exposure and pesticide residues. EFSA’s opinion.

The Prometheus Paradox

TFA is a compelling test case for EU food-safety governance. It meets the broad PFAS definition and is a “forever chemical”: extremely persistent, highly mobile and now widespread in water and the wider environment. It is also formed when certain fluorinated pesticides, refrigerants and other substances break down.

Its story captures an uncomfortable paradox of innovation. Fluorinated technologies were engineered into modern life to solve real problems -from replacing ozone-depleting refrigerants to making clothing waterproof and cookware non-stick. Not all these products generate TFA directly, but together they illustrate the vast fluorochemical economy from which today’s contamination problems emerged.

We engineered these substances into our lives to manage other risks: until their persistence became a risk in itself.

Where Hazard Meets Risk: The Three-Body Problem

Read together, these two opinions do something more than add up. They operate on entirely different logics. EFSA’s HBGVs are risk-based: they set reference points and ask, how much exposure is safe? RAC’s Repr. 1B recommendation is hazard-based: it asks, is this substance intrinsically dangerous?, and answers by category, not by dose.

That distinction matters enormously, because a Repr. 1B classification is a cut-off criterion under the Plant Protection Products Regulation (1107/2009). Active substances that fall into that category can be refused approval or renewal on hazard grounds alone, largely irrespective of the exposure calculus EFSA’s HBGVs describe.

In other words, the toxicology and the hazard classification reach the same substance through two different legal doors -and the hazard door is the harder one to keep shut.

This is where the environmental and human-health dimensions stop being separable. Until now, the case for restricting TFA leaned on persistence and groundwater, an environmental argument that implied a health concern. The RAC opinion inverts that: the human-health concern is now direct and free-standing, and it flows back into the environmental and residue framework. Protecting groundwater was always about protecting people; the reprotoxicity classification makes that literal.

The Bridge into the Omnibus Food and Feed Safety Debate

These considerations reopen the MRL question with far more force. In the Omnibus X discussions, the Council has proposed lowering MRLs - potentially to the limit of quantification (LOQ) -for non-approved active substances with particularly hazardous properties or whose use may result in unacceptable concentrations of active substances or metabolites in groundwater. The proposal also addresses substances with unacceptable effects on bees and remains subject to political debate in the ordinary legislative process.

A Repr. 1B classification hands that proposal a second, independent rationale, no longer only “persistent contaminant in water,” but “reprotoxic hazard in the food chain.” Two justifications, converging on the same number: LOQ.

EFSA’s new HBGVs provide a parallel, risk-based route to possible MRL action. They do not lower MRLs automatically, but existing residue data and proposed limits may now need to be tested against the lower Acceptable Daily Intake and the newly established Acute Reference Dose.

While Toxicological Reference Values as those released by EFSA are ny no means an indication of a lowering of MRLs, usually the establishment of MRLs requires TRVs. If estimated dietary exposure exceeds those benchmarks, risk managers may need to lower individual MRLs, amend residue definitions, request further data or restrict the uses producing TFA residues. The issue for the EU is virtual and hypothetical. No TFA releasing molecules are in the pipeline. No one has interest in setting MRLs when there are other measures to tackle TFA presence. However, for Third countries, with still many TFA substances in use, this could be a game changer and heavily affect MRL, and licence to export to the EU block.

But the balance still deserves honesty. A hazard classification is not a risk finding—the reprotoxic effects were observed in animal studies at doses above typical environmental exposure levels, and the RAC opinion is a recommendation that the Commission has yet to enact.

Precaution and proportionality are both legitimate, and they pull in different directions here.

Absence of Evidence is Not Evidence of Absence

There is a third door in this file, and it is the quietest one. Alongside the dose and the hazard sits a prior question: what would it take to know that TFA is formed at all.

At the SCoPAFF Phytosanitary Legislation meeting of 29-30 June 2026, the Commission reported back on the ad hoc working group convened under the second EFSA mandate on TFA (List – Methods – Fate). The group’s diagnosis of the literature was blunt: most of it is descriptive. It tells us where TFA has been found and in what quantities, and comparatively little about how it got there. Occurrence and monitoring data are abundant; mechanistic understanding of formation pathways is thin.

Monitoring remains indispensable, the group agreed. But it answers a different question from the one regulators increasingly need answered: not whether TFA present, but how, under what conditions, and from which precursors it is formed. Without that, attributing environmental TFA to particular active substances stays inference rather than demonstration – and attribution is precisely what cut-off criteria and any future MRL action would have to rest on.

Two responses were discussed.

The first is to borrow: whether broader PFAS research on photochemical transformation and plant uptake can fill mechanistic gaps where TFA-specific evidence does not exist.

The second is to integrate: to combine physicochemical properties, degradation behaviour and existing regulatory information into a coherent framework capable of supporting modelling, rather than leaving isolated data points to speak for themselves.

Then came the uncomfortable part. The group turned to the instruments themselves and found them wanting. Current OECD test guidelines may simply not be built to detect TFA formation: study durations may be too short, radiolabelling strategies may not track the relevant fragment, and the design may be unable to capture metabolites occurring at very low levels – or metabolites never identified in the first place. The analytical chemistry compounds the problem: contamination risk (TFA is everywhere, including in the laboratory), inter-study variability, insufficient sensitivity, and losses during sample preparation. Behind all of it sits an unresolved regulatory choice: whether assessment should concern itself only with major metabolites, or also with metabolites that matter despite occurring at low levels.

The conclusion deserves to be stated plainly: non-detection of TFA cannot be taken as evidence that no formation occurs. A negative result may describe the world, or it may describe the limits of the guideline, the label, the instrument and the blank.

This is not a technicality. It is an ontological question wearing a laboratory coat. When a dossier reports “not detected”, it makes a claim about the substance; but the claim is only ever as strong as the apparatus that produced it. If the apparatus was not designed to find the thing, its silence is not testimony.

Which is exactly the terrain the precautionary principle was written for. Precaution is not, at bottom, a preference for stringency; it is a rule for allocating the burden of proof under acknowledged uncertainty. Where the methods are known to be insufficient, the absence of a finding neither discharges the applicant’s burden nor licenses regulatory inaction.

And it circles back sharply to the MRL debate. Setting a limit at the limit of quantification means setting it at the lowest level a laboratory can reliably measure. That is a threshold defined by the method, not by the residue: it says what analysts can currently see, not what is toxicologically safe or what is actually present in the commodity. It also moves — when analytical sensitivity improves, the limit falls with it, and consignments that complied yesterday may not comply tomorrow. That is a demanding role for a number to play at precisely the moment the working group is documenting how unreliable TFA detection can be.

So there are not two logics at work in the TFA file, but three: the dose, the hazard, and the method. EFSA sets the reference values. ECHA characterises the intrinsic property. And the ad hoc working group is quietly asking whether the evidence base can bear the weight that either of them will place on it.

Policy Choices - and Political Decisions

And here is the part rarely said out loud: the evidence decides less than the framework’s choice of what counts as evidence. Whether risk logic or hazard logic governs is not something toxicology can settle - it is a political decision about the standard and the burden of proof.

The science is indispensable, but it does not choose the lens through which it becomes policy.

The strategic centre of gravity, meanwhile, remains where it was: imports. Inside the EU, relevant active substances are under review or potentially exposed to the cut-off criteria. The sharper question is third-country production, where TFA-generating crop-protection products may remain in use - and where EU MRL decisions become instruments of trade and market access as much as food safety.

EFSA has factored in the dose. ECHA has concluded on the hazard. Yet the evidence base cannot reliably establish when, how or from which precursors TFA is formed—and non-detection may reflect the limits of the method rather than absence. That adds a layer of radical uncertainty to the choice of which logic governs and how the burden of proof is allocated. Neither agency can make that choice. It belongs to the risk managers and is, unavoidably, political.