Ethylene Oxide and 2‐Chloroethanol in Foods: Hazards, Detection Markers, and Regulatory Perspectives

ABSTRACT

Ethylene oxide (EO) has re‐emerged as a major food safety concern, with findings in spices, sesame, dried vegetables, food additives, and composite foods revealing persistent gaps between toxicological hazard, analytical detectability, and regulatory interpretation. In current food control practice, 2‐chloroethanol (2‐CE) is often the dominant analyte because parent EO is highly volatile and reactive; however, the scientific basis for treating 2‐CE as a surrogate for EO remains contested. This review examines EO‐related residues through three integrated lenses: the chemistry and toxicological significance of EO and its reaction products, the analytical determination and interpretation of residues across food matrices, and the international regulatory divergence that hinders harmonization. Evidence indicates that EO itself is a compound of clear toxicological concern, whereas 2‐CE is better understood as a policy‐relevant marker of EO‐related chemistry whose meaning depends on matrix context, processing history, and regulatory purpose. Processed and composite foods intensify this challenge by weakening the link between analytical findings and source attribution. Overall, EO is not simply a residue monitoring issue but an interpretive problem created by the misalignment of hazard assessment, marker‐based analytics, and regulatory decision‐making. Progress will require clearer toxicological positioning of 2‐CE, matrix‐specific analytical reporting, structured source attribution frameworks, and closer alignment between national control systems and international standard‐setting efforts.

1. Introduction

Ethylene oxide (EO) residues have become a central concern in food safety, situated at the intersection of analytical chemistry, trade compliance, and risk communication. Once regarded primarily as a fumigant or toxicant, EO now represents a broader governance issue because a single residue finding can trigger rapid decisions by laboratories, import control authorities, retailers, and multinational food business operators. Consequently, even minor ambiguities in interpretation can produce disproportionately large downstream effects.

EO has re‐emerged as a major food safety issue because recent incidents involving sesame seeds, spices, dried vegetables, food additives, and composite foods have revealed a persistent mismatch between toxicological hazard, analytical detectability, and regulatory interpretation. Although EO is recognized as a highly reactive epoxide of toxicological concern, contemporary food control practice often relies less on direct measurement of the parent compound than on detection of 2‐chloroethanol (2‐CE), a more stable reaction product commonly interpreted as evidence of EO‐related chemistry within the supply chain (EFSA ²⁰²²).

The practical significance of this problem was highlighted during the 2021 European EO incident, in which findings in imported sesame and related products triggered extensive recalls and downstream control actions across diverse food categories. These events demonstrated that EO‐related governance cannot be understood solely in terms of raw commodity testing. Once affected materials enter stabilizers, gums, spice blends, bakery products, supplements, or other processed and composite foods, the relationship between the original source event and the final analytical finding becomes increasingly difficult to reconstruct. The result is a regulatory environment in which a single upstream residue event may generate multiple downstream decisions with varying evidentiary strength (European Commission ²⁰²¹).

A further complication is that several distinct scientific questions are often treated as interchangeable. One question asks whether EO is hazardous; the answer is well established. Another asks whether 2‐CE is a practical marker of EO‐related chemistry; in many analytical contexts, the answer is yes. A third asks whether 2‐CE should be treated as toxicologically equivalent to EO in regulatory decision‐making; this point remains contested (O'Keeffe et al. ²⁰²²). When these questions are compressed into a single residue definition problem, scientific nuance is lost and policy debate becomes circular.

The contemporary EO challenge is therefore not simply a matter of improving analytical sensitivity. Laboratories can already generate technically robust measurements. The harder question is how those measurements should be interpreted. A marker‐based result may be analytically sound but mechanistically incomplete, particularly when parent EO is absent, when the matrix has undergone extensive processing, or when the supply chain includes multiple stages of blending and redistribution. Under these conditions, the same concentration value may function as a screening flag, a compliance trigger, or an incomplete clue to historical exposure, depending on the regulatory framework in which it is applied.

This distinction has important implications for review design. A useful review of EO and 2‐CE cannot remain confined to toxicology, analytical chemistry, or regulation in isolation. It must instead connect transformation chemistry, marker‐based determination, matrix constraints, and the diverse ways authorities translate evidence into action. The main value of such a review lies not in listing incidents or methods but in showing how these domains interact and where their underlying assumptions diverge. This integrative perspective is intended for three audiences that today draw on largely separate literatures: regulators deciding on proportionate action, analytical laboratories reporting marker‐based findings, and food business operators managing cross‐border compliance. Its value lies in clarifying how their assumptions interact—an interaction that no single‐discipline account makes visible.

The present review adopts that integrative perspective. It examines EO‐related residues through three linked lenses: the chemistry and toxicological significance of EO and its reaction products; the analytical determination and interpretation of residues across food matrices; and the international regulatory divergence that continues to hinder harmonization. Emphasis is placed on separating parent compound hazard from marker‐based evidence, distinguishing analytical detection from source attribution, and clarifying why processed foods intensify the interpretive challenge.

This organizing logic also addresses a gap in existing discussions. Many practical documents focus on methods, thresholds, or incident handling, whereas individual scientific papers typically address only one part of the problem, such as toxicology, analytical performance, or occurrence data. The gap between these domains remains less clearly articulated. This gap is precisely where disagreements about EO and 2‐CE persist.

Accordingly, this review argues that EO‐related food safety is best understood as an interpretive problem created by the gap between parent compound hazard, marker‐based analytics, matrix‐dependent residue behavior, and fragmented decision‐making across jurisdictions. To develop this argument coherently, the review first examines the chemical and toxicological basis of EO‐related residues, then evaluates analytical determination and interpretation across food matrices, and finally discusses the regulatory divergence that continues to constrain international harmonization.

2. Regulatory Background: Divergent Principles and the International Landscape

2.1. Divergent Regulatory Treatment of EO and 2‐CE

One reason EO remains contentious is that jurisdictions do not begin from the same regulatory assumptions. In the European context, EO‐related control is shaped by prohibition‐oriented pesticide regulation, rapid alert coordination, and a precautionary approach to residues associated with a genotoxic carcinogen. During the European EO incident, the regulatory challenge quickly expanded from raw sesame and spice materials to food additives and a wide range of downstream foods, illustrating how a marker‐based contamination problem can cascade through an integrated food market (European Commission ²⁰²¹).

This regulatory stance reflects concern not only with hazard but also with regulatory consistency. Once a prohibited substance or a closely linked marker becomes central to crisis management, authorities face strong pressure to ensure that similar findings are handled in similar ways across products and member states. Such pressure encourages harmonized enforcement language, but it can also compress scientific uncertainty into administratively simpler categories.

Other settings have emphasized different priorities. Fowles et al., for example, assessed EO residues in imported spices through quantitative dietary cancer risk estimation and concluded that lifetime excess cancer risks were very low, while also acknowledging substantial toxicological data gaps for chlorohydrins and bromohydrins via the oral route. This does not negate the legitimacy of precautionary approaches, but it does show that risk‐oriented and hazard‐oriented frameworks may yield different practical interpretations even when they address related residue findings (Fowles et al. 2001).

The result is not simply regulatory diversity, but regulatory divergence in the meaning assigned to analytical evidence. In one setting, marker detection may function as sufficient grounds for strict enforcement. In another, the same finding may be interpreted through exposure context, route‐specific uncertainty, or case‐specific risk characterization. For food business operators and laboratories working internationally, analytical comparability therefore does not guarantee regulatory comparability (Fisher 2019).

These differences also have commercial consequences: food business operators must maintain different compliance strategies for the same ingredient across markets—an operational burden whose structural drivers are examined in Section 6.2.

2.2. Hazard‐Led and Risk‐Based Approaches, and Divergent Residue Definitions

A central reason why jurisdictions reach different regulatory conclusions on EO is that they begin from different regulatory logics. In the European Union (EU), the approach in this case is best described as predominantly hazard led. EO is no longer approved as a pesticide in the EU and has a harmonized classification as mutagenic, carcinogenic, and toxic for reproduction (all Category 1B). In the EU enforcement context, compliance has therefore been driven primarily by the detection of residues under the applicable residue definition and by limit of quantification (LOQ)‐based control values, rather than by the derivation of an acceptable dietary exposure threshold for routine risk management (ECHA n.d.; European Commission ²⁰²¹).

By contrast, the United States regulates EO within a predominantly risk‐based framework. In EPA's 2020 draft registration‐review human health assessment, the principal concern identified for the parent compound was chronic inhalation exposure, particularly cancer risk. For dietary exposure, however, EPA concluded that parent EO is highly volatile and is not expected to remain on treated spices, herbs, or dried vegetables at meaningful levels by the time of consumption; accordingly, EPA did not conduct a quantitative dietary risk assessment for parent EO (EPA ²⁰²⁰).

These different regulatory logics are reflected in divergent residue definitions. In the EU, the operational residue definition used in enforcement is the sum of EO and 2‐CE, expressed as EO, with 2‐CE converted into EO equivalents. This means that findings of the parent compound and its marker are merged into a single regulated quantity for compliance purposes (European Commission ²⁰²¹).

In the United States, by contrast, EPA establishes separate tolerances for EO and for its reaction product 2‐CE (ethylene chlorohydrin) under 40 CFR § 180.151 (Code of Federal Regulations ²⁰²⁴). The two analytes are therefore reported and regulated as distinct entities rather than combined into a single residue expression. This distinction is toxicologically significant, because EPA assesses 2‐CE on its own oral toxicity endpoints for dietary exposure and does not simply assume that it inherits the parent compound's inhalation carcinogenicity (EPA ²⁰²⁰).

The practical consequence is that the same analytical result may be treated differently across jurisdictions. A result that constitutes noncompliance under an EU residue definition built around hazard‐led, LOQ‐based enforcement may not necessarily imply an appreciable dietary risk under a US‐style risk assessment framework. The divergence is therefore not merely administrative; it reflects a deeper difference in how residue findings are translated into regulatory action, and it remains a major obstacle to international harmonization (.; EPA ²⁰²⁰).

2.3. Codex Discussions and Unresolved Standard‐Setting Questions

The Codex setting is especially important because it reveals which uncertainties are local and which are genuinely global. National systems can sometimes move forward with a precautionary default even when the underlying science remains contested, but Codex‐level work requires a broader evidentiary basis because its outputs must be legible across different regulatory cultures. EO and 2‐CE remain difficult in this setting precisely because jurisdictions are not only examining different data but also organizing the same data within divergent interpretive frames. Until those frames become more comparable, technical debate is likely to continue circling around residue definitions and occurrence data without fully resolving the deeper problem of meaning.

The Codex process illustrates why harmonization has remained difficult despite broad recognition of the issue. Recent Codex materials indicate that EO and 2‐CE have been discussed as contaminants of concern, but progress has been constrained by unresolved questions about residue definition, toxicological interpretation, data sufficiency, and the most appropriate route for international standard setting. The continuation of discussion without a clear harmonized outcome reflects the scientific complexity of the topic rather than a simple administrative delay (Codex Alimentarius Commission ^{2024a, 2024b}).

Committee‐level discussions have made clear that several questions remain open simultaneously: whether EO and 2‐CE should be addressed primarily through contaminant frameworks or pesticide‐residue logic; how occurrence data should be generated and compared across matrices; whether current toxicological assumptions are robust enough for international consensus; and how analytical results should be expressed when laboratories rely on conversion factors or marker‐based definitions. When each of these questions remains unsettled, consensus becomes structurally difficult.

This helps explain why calls for harmonization often move faster rhetorically than technically. Stakeholders can agree that inconsistent interpretation is problematic, yet still disagree on the evidentiary architecture needed to resolve it. One party may prioritize broader occurrence monitoring, another compound‐specific toxicology, another analytical method comparability, and another the legal framing of the issue. Because EO‐related governance sits at the intersection of these priorities, Codex discussion tends to expose their differences rather than quickly resolve them.

The pace of international standard‐setting is therefore not only a matter of committee procedure. It also reflects the difficulty of producing a shared language for residues whose measurement, toxicological status, and regulatory meaning are not perfectly aligned. In this respect, EO resembles a class of food control issues in which harmonization depends as much on conceptual clarification as on the generation of additional data.

A recurring source of confusion in EO incident management is the distinction between product recall and withdrawal, which carry different operational and evidentiary implications. Under EU food law, Article 19 of Regulation (EC) No 178/2002 requires food business operators to initiate withdrawal where a noncompliant food has left the immediate control of the operator, and, where the product may have reached the consumer, to inform consumers and, if necessary, recall products already supplied to them (European Commission ²⁰⁰²). In practice, this means that a contaminated ingredient or additive may be withdrawn from further use, whereas a finished product that has already reached consumers may need to be recalled. Because withdrawal generally operates earlier in the supply chain, while recall applies where consumer exposure is already possible, the choice between them should be proportionate to both the strength of the evidence and the likely exposure—an issue developed in the source attribution framework of Section 5.3.

3. EO‐Related Chemistry, Toxicology, and the Regulatory Role of 2‐CE

3.1. Chemical Behavior and Residue Pathways of EO

Because EO is highly reactive, residue interpretation is inseparable from time. A laboratory result obtained close to the point of production may differ substantially from one obtained after shipment, storage, or incorporation into a final product. Thus, the loss and conversion of EO obscure the timing of the original exposure. In practical terms, the same commodity may present different analyte profiles at different stages of the supply chain, even when the original exposure history is identical. This temporal instability is one reason why retrospective inference from final product residues should be framed cautiously in both scientific papers and regulatory communications (World Health Organization ¹⁹⁸⁵; Scudamore and Heuser ¹⁹⁷¹).

The interpretive challenge becomes greater when residue chemistry is embedded in heterogeneous food matrices. In spices and dried vegetable materials, chloride availability, botanical composition, and dehydration conditions may favor one pattern of transformation, whereas in gums, composite foods, or processed formulations, the final analyte profile may be influenced by multiple inputs and reformulations. The composition of the food matrix actively shapes which analytes later become detectable. Its composition helps determine which analytes later become detectable. A chemically informed regulatory framework should therefore distinguish between marker persistence and proof of a unique source mechanism (Tateo and Bononi 2006; Bononi et al. ²⁰¹⁴).

EO is a small cyclic ether characterized by high volatility and strong electrophilic reactivity. These properties explain both its historic use as a fumigant or sterilant and the analytical difficulty of identifying it directly in foods after storage, processing, or transport. Because EO readily undergoes ring‐opening reactions with nucleophiles such as water, halides, alcohols, and sulfhydryl‐containing compounds, the parent compound may disappear while more stable reaction products remain detectable. In food matrices, this means that a negative result for parent EO does not demonstrate that EO‐related chemistry never occurred (World Health Organization ¹⁹⁸⁵).

The relevance of this chemistry to food control is longstanding. Early work on fumigated commodities showed that reaction products can persist even when the parent gas is no longer measurable. Wesley et al. demonstrated that chlorohydrins can be formed in foodstuffs following EO fumigation, thereby establishing a durable conceptual link between processing history and persistent residues (Wesley et al. 1965). Scudamore and Heuser later reinforced this point by showing that reaction products may remain in flour and related commodities after fumigation or sterilization, and may also be affected by subsequent processing steps (Scudamore and Heuser 1971).

Among the transformation products, 2‐CE has become the dominant regulatory marker because chloride‐containing food matrices can facilitate its formation and because it is analytically more persistent than EO itself. Historical studies of fumigated foods showed that chlorohydrins can persist after EO treatment and may therefore serve as practical indicators of earlier exposure. This logic still underpins many modern monitoring programs, especially where laboratories require an operationally robust analyte for routine surveillance of imported spices, dried vegetables, and food ingredients (Tateo and Bononi 2006).

At the same time, the interpretive role of 2‐CE is best understood in relation to current regulatory and industrial practice. Under a given regulatory principle—hazard‐based or risk‐based—the residue definition determines whether 2‐CE is counted as part of the regulated residue, and that classification in turn sets the route for follow‐up action. Information about when and under what conditions the residue formed—direct fumigation, ingredient carryover, or transformation during storage and handling—does not ordinarily change this classification; rather, it supports root‐cause identification and mitigation once a finding has been made. Marker detectability and source attribution thus remain related but distinct: the former drives the compliance decision, while the latter informs corrective and preventive action (Bononi et al. 2014).

This distinction becomes especially important when residue findings are transferred from scientific interpretation into compliance decisions. A regulatory system may treat a persistent marker as sufficient evidence that a prohibited EO‐related event occurred in the supply chain. A scientific interpretation, however, is more cautious. It recognizes that the measured residue profile reflects both the original exposure and the post‐exposure history of the matrix, including moisture, salt content, temperature, storage duration, and processing severity.

Other EO‐related analytes, including 2‐bromoethanol (2‐BE), ethylene glycol (EG), and broader glycol‐related products, further demonstrate that EO cannot be reduced to a simple parent‐versus‐marker model. These compounds broaden the possible residue network and suggest that observed analyte profiles may reflect matrix composition, moisture, halide availability, and processing history rather than a single uniform pathway. Thus, a chemically informed interpretation requires not only measurement but also explicit recognition that different analytes carry different evidentiary weights (Figure 1).

FIGURE 1.

Chemical and interpretive pathways linking ethylene oxide to marker compounds. Depicted are the transformations through which EO is converted into more stable analytes that form the practical basis for food control decisions. Owing to the volatility and reactivity of EO, 2‐CE predominates as the analytical marker across many food matrices. Evidentiary strength varies by compound: parent EO constitutes direct evidence of presence, whereas 2‐CE and related derivatives are indirect markers whose interpretation requires consideration of matrix composition and product history. Source: Prepared by the authors based on the literature cited in the text.

3.2. Toxicological Concern for EO and Interpretive Uncertainty for 2‐CE

This distinction matters because regulatory language often travels beyond the expert settings in which it was originally formulated. A conversion factor or residue definition introduced for enforcement convenience can gradually be interpreted by nonspecialists as a direct statement of toxicological equivalence. Once that occurs, scientific nuance becomes difficult to recover. The problem is not that conservative policy tools are illegitimate, but that they must be described transparently. A rigorous review should therefore distinguish between a precautionary default used to avoid underestimating risk and a definitive claim that two compounds share the same biological identity. Failure to preserve that distinction has contributed materially to confusion in the EO literature and in public‐facing incident narratives (EFSA ²⁰²²; O'Keeffe et al. ²⁰²²). In this use, 2‐CE functions as a proxy: its detection licenses an inference that an EO‐related event occurred upstream, but it does not by itself establish the provenance of that event. Treating the proxy as if it recorded provenance conflates biological identity—what the compound is—with causal history—how and where it arose. The framework in Section 5.3 is designed to keep these two questions separate.

The same problem affects communication among laboratories, industry, and regulators in different jurisdictions. When 2‐CE results are reported as EO equivalents without explanation, downstream users may assume that the analytical output simultaneously establishes an exposure scenario, a toxicological conclusion, and a compliance outcome. In reality, these are separate layers of interpretation. Better toxicological positioning of 2‐CE would therefore do more than refine hazard assessment; it would also reduce ambiguity in how analytical findings are communicated and applied across food systems. This benefit is especially important in incident management, where compressed timelines often encourage oversimplified narratives (Fowles et al. 2001; EFSA ²⁰²²).

The toxicological basis for concern about EO itself is strong. The World Health Organization (WHO) Environmental Health Criteria report described EO as a highly reactive compound capable of alkylating proteins and DNA, producing mutagenic and carcinogenic effects in experimental systems, and raising concerns in human exposure settings. This toxicological profile explains why regulators often approach EO‐related residues with a high degree of precaution, even when direct dietary data are incomplete. From a hazard perspective, EO is not a marginal contaminant but a substance of longstanding toxicological significance (World Health Organization ¹⁹⁸⁵). This assessment is shared across authoritative bodies: the International Agency for Research on Cancer classifies EO as carcinogenic to humans (Group 1), the US National Toxicology Program lists it as known to be a human carcinogen, and the European Food Safety Authority (EFSA) likewise treats it as a genotoxic carcinogen without a safe threshold (IARC ²⁰¹²; NTP ²⁰²¹; EFSA ²⁰²²).

This concern is reflected in the older toxicological literature cited by risk assessors. EO has repeatedly been characterized as a direct‐acting alkylating agent, and this mechanism is central to its regulatory identity. The conceptual bridge from mechanistic toxicology to food governance is therefore relatively straightforward for EO itself: a highly reactive genotoxic compound should be controlled stringently. The difficulty arises when that regulatory logic is applied to a derivative or marker whose biological profile may not fully mirror that of the parent compound.

The scientific challenge emerges when this hazard characterization is extended from EO to 2‐CE. The EFSA concluded that the genotoxic and carcinogenic potency of 2‐CE is unlikely to exceed that of EO and retained 2‐CE within the EO residue definition for regulatory purposes. This position provides a practical framework for converting 2‐CE findings into EO‐equivalent values and for supporting precautionary regulatory action when direct EO measurement is analytically unreliable. In enforcement terms, the approach offers operational clarity. In scientific terms, however, it remains as much a risk management judgment as a compound‐specific toxicological conclusion (MFDS ^2021b).

This distinction becomes clearer when compared with the position advanced by O'Keeffe et al., who reviewed structure–activity relationships and in vitro evidence and concluded that 2‐CE should not be assumed to be a genotoxic carcinogen equivalent to EO. They argued instead that 2‐CE should be interpreted on the basis of noncancer endpoints and that a health‐protective reference dose should be derived from compound‐specific evidence. Their analysis does not eliminate concern about 2‐CE, but it challenges the practice of treating a marker compound as though it automatically inherits the full toxicological identity of its precursor (O'Keeffe et al. ²⁰²²). More recently, BfR (2026), drawing on new in vitro genotoxicity data generated under current OECD test guidelines, reached a broadly similar conclusion, judging biologically significant mutagenic activity of 2‐CE at residue‐relevant exposures to be unlikely and proposing provisional health‐based reference values for the compound (BfR ²⁰²⁶).

The disagreement here is not trivial. It reflects two different regulatory philosophies. One philosophy prioritizes caution where evidence is incomplete and accepts the use of conservative assumptions when public health protection is at stake. The other emphasizes compound‐specific evidence and resists equivalence arguments unless toxicological comparability is directly demonstrated. Both approaches have internal logic, but they do not produce identical interpretations of a 2‐CE result.

The dietary risk literature further illustrates the distinction between hazard characterization and exposure interpretation. Fowles et al. estimated very low lifetime excess cancer risks from EO residues in spices imported into New Zealand under conservative assumptions, while also noting substantial data gaps for chlorohydrins and bromohydrins by the oral route (Fowles et al. 2001). Although their study does not settle the current debate, it highlights an enduring point: the toxicological confidence associated with EO itself is not automatically transferable to every related marker compound and every food exposure scenario.

For food safety governance, the key implication is that “2‐CE as an analytical marker” and “2‐CE as a toxicological surrogate” are not identical propositions. A scientifically rigorous review must keep them separate. If they are merged without qualification, analytical convenience may be mistaken for toxicological equivalence, and regulatory certainty may appear stronger than the evidence actually supports. This is one of the main reasons why EO‐related controversy persists even when laboratories report numerically robust results.

A balanced review should acknowledge and preserve both sides of the issue. On the one hand, authorities are justified in taking marker‐based results seriously, particularly when the parent compound is unstable and the public health context favors precaution. On the other hand, scientific interpretation should resist presenting regulatory conversion factors as though they were compound‐specific toxicological truths. Recognizing such differences is essential if future harmonization efforts are to be credible.

3.3. Non‐Fumigant and Natural Sources of EO and 2‐CE

The regulatory interpretation of EO‐related residues has often proceeded on the basis that a positive finding may indicate EO treatment somewhere in the supply chain. While this interpretation is plausible in many incident settings, it is not universally straightforward: both EO and, in particular, 2‐CE may also arise through routes not directly involving deliberate fumigation, and these possibilities are relevant when interpreting marker findings (EURL‐SRM ²⁰²⁰; Bessaire and Delatour ²⁰²⁶).

EURL‐SRM has noted that 2‐CE need not originate exclusively from EO fumigation. Ethylene is endogenously produced in plants, and EO may theoretically be formed through enzymatic oxidation of ethylene. The report further notes that ethylene could theoretically react with hypochlorite in chlorinated water to form 2‐CE during processing steps such as soaking or washing, or potentially during irrigation. It also identifies other conceivable contributors to low‐level findings, including combustion‐related contamination, precursor chemicals, and cross‐contamination during processing, storage, or transport. Accordingly, a low‐level 2‐CE finding, especially near LOQ, should not automatically be treated as unambiguous evidence of EO fumigation (EURL‐SRM ²⁰²⁰).

The chemistry linking EO and 2‐CE further complicates interpretation. Because EO is highly volatile and analytically challenging, residue interpretation often depends heavily on 2‐CE as a marker. Recent analyses have argued that this situation, together with the merged EU residue definition, creates regulatory ambiguity and complicates the reliable attribution of 2‐CE findings to deliberate EO use (Bessaire and Delatour 2026). These considerations do not negate the value of monitoring, but they reinforce the need to interpret marker findings in light of matrix, processing history, and plausible alternative sources, a theme developed further in the source attribution framework in Section 5.3 (Oyeyinka et al. 2025).

4. Occurrence and Dietary Exposure

The regulatory principles and toxicological considerations discussed above acquire practical significance only insofar as EO‐related residues actually occur in foods and can give rise to dietary exposure. This section therefore synthesizes published occurrence and exposure information from incident surveillance, analytical observation reports, and regulatory assessments rather than presenting new measurements.

4.1. Occurrence Across Food Matrices

The scale of occurrence became most apparent during the EU incident that began in 2020. In the early phase of the event, the EURL‐SRM analytical observations report noted that, by November 20, 2020, roughly 140 notifications concerning EO in sesame from India had been recorded in the RASFF portal. Official EU annual reporting subsequently documented 468 RASFF notifications related to EO in 2021 and 186 notifications concerning EO and 2‐CE in 2022. Although these figures are not strictly cumulative, because the 2022 total groups EO and 2‐CE together, they clearly indicate that the issue rapidly developed from an isolated sesame seed incident into a sustained regulatory and surveillance problem affecting multiple product categories (EURL‐SRM ²⁰²⁰; European Commission ^{2022a, 2023a}).

To place this surveillance on a more systematic footing, the EU‐coordinated multiannual control program has required Member States to take and analyze random samples of specified commodities for EO according to a commodity‐by‐year schedule. Under Commission Implementing Regulation (EU) 2022/741, EO was to be analyzed in beans (dried), rye, and rice in 2023, and under Commission Implementing Regulation (EU) 2023/731, in wheat in 2024, barley and oats in 2025, and beans (dried), rye, and brown rice in 2026. This has created an ongoing, routine monitoring framework that extends beyond the original incident response (European Commission ^{2022b, 2023b}).

Published concentration data also show substantial variation, but the currently verified evidence should be described cautiously. In the EURL‐SRM (2020) observations report, the ethylene‐oxide‐related levels encountered in official sesame samples mostly ranged between 0.1 and 10 mg/kg, all above the applicable EU MRL for sesame. In addition, various sesame or sesame‐containing products obtained from local groceries showed 2‐CE residues at low levels between approximately 0.01 mg/kg (semiquantitative) and 0.1 mg/kg, or no detectable residues. These data nevertheless remain difficult to compare directly across studies and jurisdictions, because results may be reported as parent EO, as 2‐CE, or as EO equivalents under different residue definitions and enforcement logics (Section 2.2).

4.2. Dietary Exposure and Reference Values

Translating occurrence data into dietary risk has historically been difficult because EO has been treated in European regulatory practice as a genotoxic carcinogen for which no acceptable threshold for routine dietary risk management has been established. As a result, control has relied principally on residue definitions and LOQ‐based enforcement rather than on comparison with a conventional health‐based guidance value. In the United States, EPA's 2020 draft registration‐review assessment similarly identified chronic inhalation exposure, particularly cancer risk, as the principal concern for the parent compound. For dietary exposure, however, EPA concluded that parent EO is highly volatile and is not expected to remain on treated spices, herbs, or dried vegetables at meaningful levels by the time of consumption; accordingly, EPA did not conduct a quantitative dietary risk assessment for parent EO (EPA ²⁰²⁰).

Recent developments have begun to change this picture for 2‐CE specifically. In Opinion No. 016/2026, the German Federal Institute for Risk Assessment (BfR), drawing on new in vitro genotoxicity data generated under current OECD guidelines, concluded that biologically significant mutagenic activity of 2‐CE at relevant dietary exposures is considered unlikely. On that basis, BfR proposed provisional health‐based guidance values for 2‐CE, namely, an acute reference dose (ARfD) of 0.13 mg/kg body weight and an acceptable daily intake (ADI) of 0.02 mg/kg bw/day, and further recommended initiating an EU procedure to establish separate residue definitions for EO and 2‐CE. If these proposals are taken up in EU‐level assessment, 2‐CE occurrence data could, for the first time, be interpreted against explicit exposure‐based reference values rather than only within a hazard‐led residue definition framework (BfR ²⁰²⁶).

5. Analytical Determination and Interpretation Across Food Matrices

5.1. Current Analytical Workflows for EO and 2‐CE

Analytical fitness should be defined in relation to the decision context. A surveillance method, an enforcement confirmation method, and an incident investigation method may reasonably prioritize different combinations of sensitivity, breadth, and interpretive value. In EO‐related work, this context‐specific notion of fitness is especially important because the practical question is rarely only whether a compound can be measured. Rather, it is whether the measurement helps distinguish current presence, historical transformation, or downstream carryover in a way that is meaningful for the action under consideration. Making this explicit would improve method comparison and reduce the tendency to judge all workflows by a single undifferentiated performance logic.

Analytical determination of EO‐related residues faces a fundamental challenge: the parent compound is difficult to preserve and quantify due to its volatility and reactivity. Method development has therefore increasingly focused on workflows that either stabilize interpretation through marker compounds or maximize sensitivity through temperature‐controlled handling and headspace‐based analysis. In practical terms, method selection is inseparable from analyte chemistry (Wenio et al. 2023).

Recent validated approaches include headspace gas chromatography–tandem mass spectrometry (GC–MS/MS) for EO and modified Quick, Easy, Cheap, Effective, Rugged, and Safe (QuEChERS)‐based workflows coupled with GC–MS/MS or programmed temperature vaporization GC–MS/MS for 2‐CE. The Food Safety and Standards Authority of India (FSSAI) method explicitly emphasizes sample preparation below 10°C, matrix‐matched calibration, and the feasibility of simultaneously analyzing EO and 2‐CE across diverse matrices, including oilseeds, cereals, spices, herbs, food additives, fruits, and vegetables. Likewise, Wenio et al. described a rapid GC–MS/MS method designed to operationalize the full residue definition by converting measured 2‐CE into EO‐equivalent values. Together, these developments have strengthened routine monitoring capacity (FSSAI ²⁰²³; Wenio et al. ²⁰²³). It is worth distinguishing what method validation and laboratory accreditation do and do not guarantee. Accreditation to standards such as ISO/IEC 17025, together with validated methods, assures that a measurement is reproducible, traceable, and fit for its stated analytical scope; it does not certify that the reported analyte profile reflects a particular upstream event, because source attribution depends on contextual information that lies outside the accredited measurement itself.

The methodological emphasis on low‐temperature handling is not a minor technical detail but a core condition of measurement validity. Because EO may be lost during sample preparation, holding, or transfer, pre‐analytical control becomes part of analytical integrity. In other words, the distinction between true absence and preparation‐related loss can become blurred unless the method tightly controls the handling environment. This is why protocols repeatedly stress cooling, rapid processing, and matrix‐matched calibration.

Headspace‐based approaches are especially attractive when the objective is to maximize the possibility of observing parent EO or a volatile fraction of the residue profile. Extract‐based methods, by contrast, often fit better with routine multicommodity workflows and may be more practical in laboratories already equipped for pesticide‐residue analysis. Neither approach is universally superior; their value depends on the matrix, the analyte of interest, and the interpretive purpose of the analysis. A laboratory investigating a new enforcement incident may prioritize preserving parent EO where feasible, whereas a surveillance laboratory may prioritize robust routine detection of 2‐CE across a broad matrix range.

This difference in purpose is analytically important. Methods are often described primarily in terms of LOQs, recoveries, and reproducibility, but in EO analysis, the interpretive role of the method also matters. While a workflow optimized for a stable marker may achieve high precision, it conveys a narrower account of the exposure event than a method that preserves access to parent EO. Consequently, method validation should not be conceptually separated from result interpretation.

Analytical improvement, however, does not eliminate interpretive uncertainty. A highly sensitive method may quantify an analyte with excellent precision while still leaving unresolved what the finding means mechanistically or regulatorily. For EO‐related residues, this gap is especially important because the reported value may represent parent EO, 2‐CE, or an EO‐equivalent calculation derived from 2‐CE. Unless reports clearly distinguish among these categories, numerical accuracy can create a false sense of interpretive certainty.

For the same reason, reporting language deserves more attention than it usually receives. A statement that “EO was found” may be analytically imprecise if the laboratory actually measured only 2‐CE and expressed the result as EO equivalent. Such wording may be acceptable in some regulatory contexts, but it can obscure the distinction between what was measured and what was inferred. In a controversial area such as EO, precision of language is part of scientific quality.

5.2. Matrix Effects and Method‐Dependent Interpretation

Since EO results vary across different food matrices, matrix classification should be incorporated as a formal component of EO reporting rather than treated as an informal background detail. Results from raw spices, dried herbs, food additives, oilseeds, and composite foods should not be regarded as analytically interchangeable merely because they can all be expressed in milligrams per kilogram (mg/kg). A more informative reporting format would specify not only the concentration but also the matrix class, the principal target analyte, the sample condition at receipt, and whether the result reflects direct measurement of parent EO, direct measurement of 2‐CE, or a conversion‐based residue definition. Such structured reporting would improve interpretive reproducibility even where full harmonization remains out of reach (FSSAI ²⁰²³; Wenio et al. ²⁰²³).

Matrix‐specific interpretation is also key for robust study design. Occurrence studies that aggregate unlike matrices under a single residue category may obscure meaningful differences in how marker findings arise and persist. For example, a prevalence estimate that combines dry spices, gums, and composite foods may be useful for surveillance, but far less informative for mechanistic inference or regulatory comparison. Future studies should therefore distinguish more clearly among surveillance value, mechanistic value, and policy value when presenting EO‐related datasets. Doing so would reduce the recurrent slippage between analytical observation and broad causal claims (Bononi et al. 2014; Wenio et al. ²⁰²³).

Matrix dependence is a major reason EO‐related results must be interpreted with caution, and it has both analytical and conceptual consequences. Analytically, it influences recovery, cleanup requirements, signal suppression or enhancement, and the practical choice of calibration strategy. Here, “matrix effects” refer specifically to the analytical phenomena by which co‐extracted matrix constituents alter extraction efficiency or instrumental response (signal suppression or enhancement); these are, in principle, correctable through matrix‐matched calibration or standard addition. Conceptually, it shapes what a result can plausibly mean. A 2‐CE finding in a dry spice may be interpreted differently from the same numerical value in a gum, a high‐moisture food, or a finished, ready‐to‐eat product because the pathways of formation, persistence, and carryover are not identical across these categories. It should be stressed that this is a difference in interpretation, not in measurement: once matrix effects are properly corrected through matrix‐matched calibration, an identical corrected concentration denotes an identical amount of analyte regardless of matrix. What differs across matrices is what that concentration implies about source, not the quantity measured.

The Italian survey by Bononi et al. illustrates both the strengths and limitations of targeted control. At the same time, the study also demonstrated that marker‐driven surveillance is structured by a residue definition framework rather than by direct observation of the original exposure event (Bononi et al. 2014).

This point warrants emphasis because residue definitions are often treated as analytically neutral; in fact, they function as interpretive instruments. A residue definition determines which compounds are included, how they are combined, and how the final value is understood by regulators, laboratories, and food business operators. In EO‐related analysis, the conversion of 2‐CE into an EO‐equivalent value is not merely a computational convenience; it is a policy‐laden act that shapes how the result is perceived.

Accordingly, matrix effects should be regarded as interpretive variables rather than dismissed as technical nuisances; the framework developed in Section 5.3.2 expands this argument. Table 1 summarizes selected occurrence and analytical method studies that have influenced current practice and illustrates the interpretive heterogeneity that motivates a structured framework.

TABLE 1.

Selected occurrence and analytical method studies for EO and 2‐CE in foods.

Study (year)	Region	Matrix examined	Scope	Key finding	Basis	Platform
Wesley et al. (1965)	USA	Wheat flour, corn meal, raisins, prunes	Method study	First systematic demonstration that EO fumigation produces persistent chlorohydrin residues, including 2‐CE	EO + chlorohydrins	Colorimetric/GC
Scudamore and Heuser (1971)	UK	Wheat flour and other commodities	Method and persistence	Documented persistence of EO and reaction products through processing; established baseline expectations for marker stability	EO + chlorohydrins	GC
Fowles et al. (2001)	New Zealand	Imported spices	Risk assessment	Estimated very low lifetime excess cancer risk under conservative dietary scenarios; flagged data gaps for chlorohydrins by oral route	EO	Risk modeling
Tateo and Bononi (2006)	Italy	Spices	Method development	Established 2‐CE as a practical surrogate marker for EO fumigation; provided a basis for routine surveillance	2‐CE	GC–MS
Bononi et al. (2014)	Italy	Imported herbs, spices, dried vegetables	Survey	Approximately 29% of samples exceeded 0.3 mg/kg EO equivalents (sum of EO and 2‐CE)	EO equivalents	GC–MS
RIVM (2020)	Netherlands/EU	Sesame seeds	Risk assessment	Risk assessment supporting EU response to elevated EO findings in imported sesame; informed RASFF coordinated action	EO + 2‐CE	Risk assessment
European Commission (2021)	EU (multistate)	Sesame, additives, composite foods	RASFF incident summary	880+ RASFF notifications generated by EO findings; established present‐day reference incident for marker‐based recalls	EO equivalents	Multimethod
MFDS (2021a)	Republic of Korea	Instant noodle garnishes and seasoning powders	Investigation	2‐CE detected at approximately 0.1–12 mg/kg; parent EO not detected; informed provisional national limits	2‐CE	GC–MS/MS
FSSAI (2023)	India	Oilseeds, cereals, spices, herbs, additives, fruits, vegetables	Method validation	Validated simultaneous determination of EO and 2‐CE across diverse matrices; emphasized handling below 10°C and matrix‐matched calibration	EO + 2‐CE	GC–MS/MS
Wenio et al. (2023)	Poland/EU	Multiple food matrices	Method development	Rapid GC–MS/MS method operationalizing the EU residue definition by converting measured 2‐CE into EO equivalents	EO equivalents	GC–MS/MS
Bessaire et al. (2023)	Italy/EU	Fumigated foodstuffs (model)	Mechanistic confirmation	Confirmed near‐complete conversion of EO to 2‐CE under typical fumigation conditions; supports the EU conversion factor logic but not full toxicological equivalence	EO → 2‐CE	GC–MS
Anyogu et al. (2024)	International	Sesame value chains	Narrative review	Contextualizes EO incidents within broader sesame‐associated hazards (Salmonella, mycotoxins, pesticides); highlights supply chain complexity	Review	N/A

Note: The selected studies illustrate how the analytical and regulatory landscape for EO and 2‐CE has evolved from early chlorohydrin formation work through marker‐based surveillance to mechanistic confirmation of EO‐to‐2‐CE conversion. The list is interpretive rather than a comprehensive systematic review. Quantitative comparability across studies is limited by differences in matrix scope, residue definition (EO vs. 2‐CE vs. EO equivalents), analytical platform, and reporting conventions.

Abbreviations: 2‐CE, 2‐chloroethanol; EO, ethylene oxide; EU, European Union; FSSAI, Food Safety and Standards Authority of India; GC, gas chromatography; GC–MS, gas chromatography–mass spectrometry; GC–MS/MS, gas chromatography–tandem mass spectrometry; MFDS, Ministry of Food and Drug Safety (Republic of Korea); RASFF, Rapid Alert System for Food and Feed.

Source: Compiled by the authors from the references cited in the table.

A further challenge is comparability across laboratories. Even when laboratories employ broadly similar instrumental platforms, differences in sample treatment, cleanup strategy, derivatization logic, temperature control, or reporting conventions can alter both quantitative outcomes and narrative interpretation. Thus, the need for harmonized performance criteria is accompanied by a need for harmonized interpretive descriptors. Laboratories should converge not only on how measurements are made, but also on how results are explained (Figure 2; Table 2).

FIGURE 2.

Analytical workflow and interpretation challenges for EO and 2‐CE across food matrices. The workflow proceeds from temperature‐controlled sampling through to regulatory interpretation. Method selection is matrix dependent, and interpretation varies according to whether EO, 2‐CE, or both are detected. Because chromatographic data alone cannot establish source attribution, final interpretation requires integration of analytical results with matrix composition and supply chain context. HS–GC–MS/MS, headspace gas chromatography tandem mass spectrometry; PTV–GC–MS/MS, programmed temperature vaporization gas chromatography tandem mass spectrometry; 2‐CE, 2‐chloroethanol. Source: Prepared by the authors based on the literature cited in the text.

TABLE 2.

Representative analytical approaches for EO and 2‐CE across food matrices.

Approach	Primary target	Typical matrices	Major strengths	Major limitations	Interpretive note
HS‐GC–MS/MS	EO (and in some workflows 2‐CE)	Volatile residue assessment; sesame; dry foods	Strong sensitivity for volatile analytes; reduced solvent burden	Highly dependent on sample integrity, vial conditions, and temperature control	Best suited when preservation of parent EO is analytically plausible
Modified QuEChERS + GC–MS/MS	2‐CE and EO‐related residue definition	Spices, herbs, cereals, food additives, fresh produce	Broad applicability across diverse matrices; practical for routine laboratories	Matrix effects and cleanup burden vary considerably by product type	Commonly yields marker‐centered results requiring contextual interpretation
PTV–GC–MS/MS	2‐CE with improved injection control	Complex dry matrices; processed foods	Improved handling of extracts and sensitivity in some workflows	Instrumentation and optimization requirements may reduce comparability between laboratories	Useful when extract‐based workflows are required
Marker‐only determination via EO‐equivalent conversion	2‐CE expressed as EO equivalent	Incident response and routine control reporting	Operationally simple for enforcement communication	May blur the distinction between analytical measurement and toxicological meaning	Should be reported transparently as a conversion‐based expression
Multianalyte contextual workflow	EO + 2‐CE + related products	Investigative or source attribution cases	Improves interpretive depth beyond single‐analyte reporting	Not always available in routine monitoring systems	Most useful when enforcement depends on causal interpretation rather than screening alone

Abbreviations: 2‐CE, 2‐chloroethanol; EO, ethylene oxide; HS–GC–MS/MS, headspace gas chromatography–tandem mass spectrometry; PTV–GC–MS/MS, programmed temperature vaporization gas chromatography–tandem mass spectrometry; QuEChERS, Quick, Easy, Cheap, Effective, Rugged, and Safe.

Source: Author‐compiled synthesis based on the literature cited in the text.

5.3. Source Attribution in Processed and Composite Foods: Toward a Structured Risk Management Framework for Residue Detection of EO and 2‐CE

5.3.1. The Unit‐of‐Interpretation Problem

Processed and composite foods highlight the limits of analyte‐based regulation more sharply than raw commodities. In a single‐ingredient raw material, the chain between sample, residue, and likely source event is already imperfect but still relatively direct. In a multi‐ingredient food, residues may be diluted, redistributed, transformed, or carried over from a minor component that contributes little to final exposure but exerts disproportionate regulatory consequences. This weakens the connection between analytical detection and straightforward enforcement logic (European Commission ²⁰²¹).

Composite foods also intensify uncertainty because they blur the unit of interpretation. A single residue finding can simultaneously refer to (i) the implicated ingredient, (ii) the finished product, (iii) the consumer exposure scenario, or (iv) the tradeable lot. In some regulatory settings, these dimensions are collapsed into a single object of action, yet analytically they remain distinct: a finished product may test positive because one ingredient carried a marker at low concentration, yet the public health significance, traceability implications, and recall proportionality may differ substantially from those of the originating raw material.

Anyogu et al. (2024) emphasized that sesame‐associated food safety challenges already include microbial hazards such as Salmonella, mycotoxins, and pesticide‐related concerns. EO‐related incidents therefore do not occur in an analytical vacuum; they arise within complex supply chains where multiple hazard control systems intersect. Once affected ingredients enter processed foods, control decisions must consider not only whether a residue is detectable but also how ingredient composition, consumer exposure, traceability, and proportionality should shape the response.

5.3.2. A Structured Source Attribution Framework

Although the limits of single‐analyte logic have been widely acknowledged, no operational framework has yet translated this recognition into routine practice. The framework outlined below organizes a single residue finding along four analytical dimensions and one inferential dimension, yielding a structured judgment about the strength of evidence the finding can support and, by extension, the regulatory response it can proportionately justify. The framework is intended as a description of good interpretive practice; it complements rather than replaces jurisdiction‐specific risk management procedures. Because these dimensions draw on partly independent lines of evidence—analytical, compositional, and documentary—their convergence functions as a form of triangulation: confidence in source attribution rests not on any single measurement but on the agreement of multiple, imperfectly correlated indicators. Although developed for EO and 2‐CE, this logic is not specific to the pair: the same four‐dimension structure could apply to other marker‐based residue problems in which a stable marker is used to infer an unstable or unobservable parent event—with the nitrofuran precedent in Section 6.1 being one candidate. We flag this generalizability here and return to it among the research priorities in Section 7.

5.3.2.1. Dimension 1—Analytical Character

The first question is what the laboratory actually measured. Parent EO detected at a quantifiable concentration is the strongest direct indicator that EO chemistry has occurred recently in the matrix. Detection of 2‐CE alone, in contrast, is consistent with EO chemistry but does not establish when, where, or how the precursor was introduced. Whether the result is reported as 2‐CE on its own basis or converted to EO equivalents under a residue definition is itself part of the analytical character; conversion‐based reporting merges two separable inferences—analytical detection of a marker and toxicological treatment of that marker—into a single number that is more legible to enforcement systems but less informative for source attribution.

5.3.2.2. Dimension 2—Matrix Class

The chemical and physical character of the matrix shapes the interpretive weight of any residue finding. Dry single‐ingredient commodities (whole spices, sesame seed, dehydrated herbs) preserve a relatively close link between residue and matrix‐specific exposure history. High‐moisture single‐ingredient commodities can shift the EO/2‐CE/glycol balance through hydrolysis but still permit reasonably direct interpretation. Multi‐ingredient processed foods (seasoning blends, bakery products, supplements containing gums or stabilizers) introduce dilution, redistribution, and carryover effects that progressively decouple the analytical signal from a specific source event. Heavily reformulated or finished composite foods (ready‐to‐eat meals, multicomponent snacks, infant formulae with multiple stabilizing agents) typically represent the weakest link between analyte and original exposure.

5.3.2.3. Dimension 3—Supply Chain and Temporal Context

Even an analytically clean result is interpretively limited without information on origin and time. Relevant inputs include documented fumigation history of the original raw material; whether the implicated ingredient came from a single supplier or from blended sources; whether the affected lot has a clear chain‐of‐custody record; the elapsed time between any documented or plausible exposure event and the analytical sampling; and whether intermediate stages of storage and reformulation could plausibly have transformed or redistributed residues. Where supply chain documentation is available, it should be used as a co‐input to interpretation, not as a separate enforcement track. The framework does not require the upstream event to have been intentional. A documented fumigation and an unintended cross‐contamination or processing‐related formation are treated alike for tier assignment, because the tiers grade the strength of evidence linking the marker to an upstream EO‐related event, not the intent behind it. Where intent can be established, it informs the subsequent enforcement response rather than the inference tier.

5.3.2.4. Dimension 4—Plausibility of Alternative Sources

A complete interpretation should also evaluate whether the marker could plausibly have arisen from sources other than EO‐related chemistry. For 2‐CE in particular, the empirical evidence for non‐EO sources in food matrices is currently limited but not zero, and the precedent reviewed in Section 6.1 cautions against assuming a single‐source interpretation by default. The documented non‐nitrofuran sources of semicarbazide illustrate that marker‐based regimes have repeatedly had to confront alternative source questions after initial residue definitions were already in regulatory use. In incident response contexts, the operational rule should be: marker presence is a probabilistic indicator of EO‐related chemistry, not a deterministic proof, and the strength of that probability depends on the four preceding dimensions.

5.3.2.5. Dimension 5—Inference Strength (Tier Classification)

The four analytical dimensions can be combined into a three‐tier inference typology, summarized in Table 3. Tier 1 (strong inference) applies when parent EO is detected, the matrix is single‐ingredient, supply chain origin is documented, and alternative sources are implausible; in such cases, attribution to a specific exposure event is well supported. Tier 2 (moderate inference) applies when only 2‐CE is detected at a meaningful concentration in a single‐ingredient or lightly processed matrix with partial supply chain traceability and no clear alternative source; in such cases, the finding supports the conclusion that EO‐related chemistry has occurred somewhere upstream, but the timing, route, and specific event remain uncertain. Tier 3 (weak inference) applies when only 2‐CE is detected, the matrix is highly composite or reformulated, supply chain documentation is incomplete, and alternative sources cannot be ruled out; in such cases, the finding remains regulatorily relevant—it can support precautionary action—but should not be treated as compelling evidence about a specific upstream event.

TABLE 3.

Tier‐of‐inference typology for EO‐related residue findings.

Tier	Analytical character (D1)	Matrix class (D2)	Supply chain context (D3)	Alternative sources (D4)	Strength of inference	Indicative response (D6)
Tier 1 (strong)	Parent EO detectable; 2‐CE supportive; reporting basis transparent	Single‐ingredient raw or lightly processed	Documented origin and timing; chain‐of‐custody intact	Implausible after structured review	A specific fumigation or exposure event can be attributed	Targeted lot‐ or supplier‐specific enforcement
Tier 2 (moderate)	2‐CE alone at a meaningful concentration; parent EO absent or not detected	Single‐ingredient or modestly processed	Partial traceability; some uncertainty in timing or supplier mix	Not characterized; assumed minor without explicit evidence	EO‐related chemistry occurred upstream; specific event unknown	Precautionary recall plus structured supplier investigation; communicate uncertainty
Tier 3 (weak)	2‐CE only; concentration low to moderate; conversion‐based reporting	Multi‐ingredient composite or finished reformulated product	Limited or fragmented; long elapsed time; multiple supplier inputs	Plausible but not characterized; alternative pathways cannot be ruled out	Marker presence consistent with EO‐related chemistry; no specific upstream event supported	Precautionary action where public health context warrants; initiate surveillance feedback rather than final causal attribution

Note: (i) Tier classification is a judgment, not a calculation; the table summarizes typical patterns but does not replace case‐specific assessment. (ii) Tier assignments may shift as supplementary information becomes available. (iii) The indicative response column is illustrative; specific actions remain the responsibility of competent authorities and may differ across jurisdictions.

Abbreviations: 2‐CE, 2‐chloroethanol; EO, ethylene oxide.

Source: Author‐compiled synthesis based on the literature cited in the text.

5.3.2.6. Dimension 6—Proportionate Regulatory Response

Inference strength should inform, though not mechanically dictate, regulatory response. Tier 1 findings can support targeted enforcement against an identified upstream actor, including ingredient‐specific or supplier‐specific recall, with high evidentiary confidence. Tier 2 findings can support precautionary recall and a structured supplier investigation, but should typically be paired with explicit acknowledgment that the upstream event has not been fully identified. Tier 3 findings can support market withdrawal action where the public health context warrants precaution, but the action should be clearly described as precautionary rather than evidentiary, and should generate a feedback loop into surveillance and supplier qualification systems rather than a final causal attribution. The goal of this graded response is to maintain consumer protection while preserving the credibility of the underlying scientific narrative—credibility that is undermined when uniformly strong enforcement language is applied to evidentially heterogeneous findings.

5.3.3. Worked Illustration

Two contrasting schematic scenarios, drawn from features common in the published incident literature, clarify how the framework operates in practice.

5.3.3.1. Scenario A

Parent EO at 0.3 mg/kg and 2‐CE at 0.6 mg/kg (sum expressed as EO equivalents above 0.05 mg/kg) are detected in a single lot of sesame seed from a documented exporter, with shipment records consistent with recent fumigation. Applying the framework: D1 (analytical character) shows both parent and marker present; D2 (matrix class) is single‐ingredient raw; D3 (supply chain context) is documented and recent; D4 (alternative sources) is implausible. The finding meets Tier 1 criteria. Targeted enforcement against the identified lot or supplier is well supported, and the regulatory narrative can attribute the residue to a specific fumigation event with high evidentiary confidence.

5.3.3.2. Scenario B

Only 2‐CE at 0.04 mg/kg is detected in a finished cookie product whose ingredient list includes sesame seed, guar gum, and a spice blend, each sourced from multiple regions through an intermediary distributor. The cookie has been on the market for several weeks, and parent EO is not detectable in any retained sample of the implicated ingredients. Applying the framework: D1 shows marker only, with conversion‐based reporting; D2 is finished composite; D3 is fragmented (multiple suppliers, intermediate distributor, several weeks elapsed); D4 cannot be fully excluded, given the absence of comprehensive ingredient‐level data. The finding meets Tier 3 criteria. Precautionary action—for example, withdrawing the affected Stock Keeping Units (SKUs) and initiating a supplier qualification review—may be justified, but regulatory communication should explicitly state that the underlying mechanism has not been fully characterized.

The two scenarios are intentionally schematic. Real‐world cases routinely fall between these poles, and the framework is most useful when applied to such intermediate cases, where it supports a transparent, reproducible judgment about how much the available evidence will bear. The framework also underscores that two findings with similar numerical concentration may carry very different evidentiary weight—a point often understated in enforcement narratives focused narrowly on threshold compliance.

5.3.4. A National Example: Korea's Provisional Rule‐Making in the Absence of Harmonized Standards

The Korean response to the 2021 European EO incident illustrates a third regulatory orientation distinct from both the EU's prohibition‐centered control and exposure‐based national assessments. In mid‐2021, European notifications involving Korean instant noodle exports prompted on‐site investigation and product testing by the Korean Ministry of Food and Drug Safety (MFDS). Although parent EO was not detected, 2‐CE was identified in dried vegetable garnishes and seasoning powders at concentrations spanning approximately 0.1–12 mg/kg across the products examined (MFDS ^2021a). At the time, no Korean food standard existed for either EO or 2‐CE; the default 0.01 mg/kg pesticide maximum residue limit (MRL) technically applied but did not provide an operationally workable framework for processed multi‐ingredient foods.

In response, the Food Hygiene Review Committee was convened in August 2021 and established provisional limits expressed as 2‐CE rather than as EO equivalents: 30 mg/kg for general agricultural, livestock, fishery, and processed foods, and 10 mg/kg for foods intended for infants and young children. MFDS subsequently issued analytical guidance, including a method for EO and 2‐CE in foods and a sesame‐specific method, both based on GC–MS/MS with matrix‐matched sample preparation (MFDS ^{2021b, 2021c}).

The Korean approach is informative for three reasons. First, it shows that a national authority faced with an emerging marker‐based contaminant problem can converge rapidly on an analytical and regulatory framework even when international consensus is absent. Second, the chosen provisional level (30 mg/kg as 2‐CE) sits between the very strict EU residue definition—in which the sum of EO and 2‐CE is expressed as EO equivalents and default LOQ‐based MRLs typically operate at 0.02–0.1 mg/kg—and the historically much higher US and Canadian MRLs applied to spices and seeds. This positioning is itself an interpretive statement: Korea did not equate 2‐CE with EO for risk management purposes but treated the marker as a separately characterizable contaminant warranting its own threshold. Third, the choice to express the limit on a 2‐CE basis rather than an EO‐equivalent basis is a deliberate analytical policy move that bypasses the conversion factor central to the EFSA framework.

This case clarifies why the present review treats the toxicological positioning of 2‐CE as a live question. Different jurisdictions can reach defensible but materially different conclusions when the underlying science permits more than one reasonable interpretation, and the Korean approach demonstrates that marker‐based regulation need not collapse into marker‐as‐surrogate framing.

5.3.5. Implications for Regulatory Practice

Adopting a tiered framework of this kind has three practical implications. First, ingredient‐level surveillance, final product interpretation, and source attribution investigation should be treated as analytically distinct tasks rather than as a single integrated enforcement workflow. Each task draws on a different combination of the dimensions described above, and combining them produces narratives in which evidentiary strength is overstated. Second, regulatory communication should explicitly identify the inference tier underlying any public action. This does not require disclosure of all underlying evidence; however, it does require acknowledging that some actions rest on stronger evidentiary foundations than others. Third, where international harmonization is sought (Section 6), shared inference tier language is more achievable in the short term than full agreement on numerical thresholds, because tier descriptors can be aligned across jurisdictions even when local risk management defaults differ.

In summary, source attribution in processed and composite foods is not a problem that improved laboratory methods alone can resolve. It is a structural feature of modern food systems that requires explicit interpretive practice—a practice the framework outlined here is intended to support. Figure 3 situates this tier‐based source attribution step within a broader stepwise policy architecture, extending from analytical confirmation through proportionate regulatory response. Subsequent sections (Section 7) develop research and policy priorities consistent with this framework, including its potential role in international harmonization efforts.

FIGURE 3.

Policy architecture for regulatory decision‐making on EO and 2‐CE in foods. Depicted is a stepwise framework for handling EO‐related findings in foods, in which analytical confirmation is followed by matrix classification, toxicological interpretation, and source attribution review prior to enforcement. National actions feed back into international harmonization efforts, particularly where residue definitions and interpretive criteria remain contested. The framework supports proportionate, scientifically transparent decision‐making in both domestic and trade‐related contexts. Source: Prepared by the authors based on the literature cited in the text.

6. Analogous Precedents and the Limits of Harmonization

6.1. Lessons From Analogous Marker‐Based Residue Precedents

The interpretive challenges associated with EO and 2‐CE are not unique to this contaminant pair. Several established regulatory frameworks already manage residues through markers because the parent compound is unstable, undetectable, or not the operationally relevant endpoint. Comparison with a well‐developed precedent—nitrofuran veterinary drugs, for which stable tissue‐bound metabolites serve as the enforcement target—clarifies which features of the EO/2‐CE problem are common to marker‐based regulation generally and which are specific to EO chemistry and to food supply chain architecture. Table 4 summarizes the comparative architecture of the two regimes.

TABLE 4.

Marker‐based residue regimes: comparative architecture.

Residue regime	Parent compound(s)	Regulated marker	Toxicological assumption	Distinctive interpretive issue
Nitrofurans (veterinary)	Furazolidone, furaltadone, nitrofurantoin, nitrofurazone	Tissue‐bound metabolites: AOZ, AMOZ, AHD, SEM	Metabolite presence taken as evidence of prohibited parent use	SEM has documented non‐nitrofuran sources, weakening source attribution
EO/2‐CE	EO	2‐CE, with limited use of 2‐bromoethanol and ethylene glycol	EU: sum of EO + 2‐CE expressed as EO equivalents. Korea: 2‐CE on its own basis. Toxicological equivalence between 2‐CE and EO is contested.	Marker is more stable than parent; conversion assumption central but disputed; non‐EO sources of 2‐CE not yet systematically characterized

Abbreviations: 2‐CE, 2‐chloroethanol; AHD, 1‐aminohydantoin; AMOZ, 3‐amino‐5‐morpholinomethyl‐2‐oxazolidinone; AOZ, 3‐amino‐2‐oxazolidinone; EO, ethylene oxide; SEM, semicarbazide.

Source: Author‐compiled synthesis based on the literature cited in the text.

6.1.1. Nitrofuran Metabolites (AOZ, AMOZ, AHD, SEM)

For the prohibited veterinary nitrofurans—furazolidone, furaltadone, nitrofurantoin, and nitrofurazone—the parent drugs deplete from edible tissue within hours of administration, while their tissue‐bound side‐chain metabolites (3‐amino‐2‐oxazolidinone, AOZ; 3‐amino‐5‐morpholinomethyl‐2‐oxazolidinone, AMOZ; 1‐aminohydantoin, AHD; and semicarbazide, SEM) persist for weeks to months and are therefore the only operationally usable enforcement target (Cooper et al. 2005; FDA ²⁰⁰⁸). The conceptual parallel with 2‐CE is striking: in both cases, the marker is more stable than the parent and is detected long after the parent is no longer present. The nitrofuran case has, however, generated controversy because semicarbazide can also arise from non‐nitrofuran sources, including azodicarbonamide‐based packaging materials and certain food‐processing reactions. This precedent reinforces the principle that marker‐based enforcement is most defensible when (i) the marker is specific or near specific to the regulated parent or (ii) interpretive rules explicitly address alternative sources.

This precedent shares a common architecture with EO/2‐CE but differs in important details. Nitrofuran metabolites illustrate the temporal advantage of stable markers and the disputes that follow when the marker is not fully specific. The same tier‐of‐inference logic developed in Section 5.3 applies here: a semicarbazide finding supports a strong inference when the marker is specific to a documented nitrofuran source, but only a weak inference when non‐nitrofuran sources—such as azodicarbonamide‐treated packaging—cannot be excluded.

For EO and 2‐CE, this precedent suggests three implications. First, the marker‐based residue definition currently applied to EO/2‐CE in the EU is firmly embedded within an established regulatory tradition; criticism of this specific application should not be misconstrued as criticism of marker‐based regulation in general. Second, the unresolved questions for 2‐CE—whether the marker has non‐EO sources, whether toxicological equivalence should be assumed by default, and how matrix effects modify interpretation—are recognizable variations of questions that have been at least partially addressed in adjacent regulatory areas. Third, the absence of a precedent‐aware framing in current EO governance is itself a missed opportunity: the EO/2‐CE debate has been conducted largely in isolation, even though adjacent residue regimes already provide a partial template for how marker, surrogate, and equivalent are best distinguished in food safety governance.

6.2. Why Harmonization Remains Difficult

Harmonization is often discussed as though it were a single endpoint, yet in practice, it encompasses multiple dimensions. These include harmonization of methods, residue definitions, toxicological assumptions, reporting language, and enforcement response. Progress in one dimension does not guarantee progress in the others. EO provides a particularly clear example of this fragmentation: laboratories may converge on analytical capability while authorities continue to diverge on what the same result should mean. Recognizing these layers can help prevent unrealistic expectations about how quickly international consensus can be achieved (Codex Alimentarius Commission ^{2024a, 2024b}).

Thus, incremental harmonization is more realistic than immediate comprehensive agreement. A first step could involve shared terminology distinguishing measured analytes from converted residue definitions. A second could be the generation of matrix‐stratified occurrence data. A third could be the adoption of a common template for expressing uncertainty around source attribution. None of these measures would resolve the toxicological debate outright, but together they would reduce the interpretive noise that currently makes comparison difficult. In policy terms, harmonization should be understood as a sequence of interoperable improvements rather than a single consensus event (Fisher 2019; Codex Alimentarius Commission ^2024a).

The deeper reason harmonization remains limited is that EO sits at the intersection of several systems that are not naturally aligned: toxicology, analytical chemistry, contaminant control, pesticide regulation, international trade, and food industry traceability. These systems pose different questions. Toxicology asks what type and level of harm are plausible. Analytical science asks what can be measured reliably. Regulators ask what action can be justified consistently. Trade systems ask whether rules are predictable across borders. EO‐related residues expose the gap among these questions rather than fitting neatly within any single framework.

Underlying these layers is the more basic divergence set out in Section 2.2: a hazard‐led regime treats residues above the applicable legal limit under the relevant residue definition for a non‐threshold carcinogen as actionable, whereas a risk‐based regime conditions action on estimated dietary exposure and applicable tolerance thresholds. The practical stakes are clearest in a common scenario. When a contaminated ingredient makes up only a small fraction of a finished product, as with some stabilizers and additives, a hazard‐led logic may still support withdrawal and, where the product has reached consumers and other measures are insufficient, recall, because the relevant residue remains legally actionable under that framework; a risk‐based logic would first ask whether the resulting dietary exposure is appreciable. The same finding can thus support withdrawal under one principle and a more limited response under the other, which is why numerical harmonization remains elusive until this difference in first principles is acknowledged (Section 2.2).

In this sense, the absence of harmonization is not merely a failure to agree on numerical thresholds. It reflects a deeper failure to align categories of interpretation. Until international bodies and national regulators more clearly distinguish parent compound hazard, marker‐based evidence, source attribution, and matrix‐dependent regulatory meaning, debates about EO and 2‐CE are likely to recur even as analytical methods continue to advance.

The broader lesson is that unresolved food safety controversies increasingly arise not because measurement is impossible, but because interpretation is socially and institutionally distributed across multiple systems. EO thus serves as a case study in how the pace of analytical innovation can outstrip consensus on meaning. This gap must be addressed explicitly if harmonization is to become more than a procedural goal (Table 5).

TABLE 5.

International regulatory orientations relevant to EO and 2‐CE in foods.

Jurisdiction/forum	Regulatory orientation	Treatment of EO/2‐CE	Practical consequence	Key implication for harmonization
European Union	Precautionary, prohibition‐centered control with coordinated incident handling	2‐CE often treated within EO residue definition logic	High sensitivity to marker findings and broad downstream recall implications	Provides strong control but can intensify trade friction when definitions differ
India (method and export control context)	Method development and trade facilitation under intensified scrutiny	Focus on validated determination of EO and 2‐CE across matrices	Supports laboratory reproducibility and export compliance	Shows how analytical standardization can become central to trade response
New Zealand (risk assessment example)	Exposure‐oriented interpretation in the dietary context	EO and related markers considered within conservative intake modeling	Illustrates low estimated dietary cancer risk under specific assumptions	Highlights the importance of route‐specific exposure assessment
Food business operator perspective	Compliance across divergent national requirements	Marker detection may trigger different responses across markets	Creates uncertainty in sourcing, recall strategy, and documentation	Demonstrates the operational cost of non‐harmonized interpretation
Republic of Korea	Provisional national standard developed through the Food Hygiene Review Committee following the 2021 incident	2‐CE expressed as 2‐CE (not EO equivalents); 30 mg/kg general foods, 10 mg/kg infant/young‐child foods	National analytical methods issued by MFDS; bypasses the EFSA conversion factor framework	Demonstrates that marker‐based regulation need not adopt the marker‐as‐surrogate logic
United States/Canada	Approval of EO as an antimicrobial fumigant for selected commodities	Historic tolerances of 7–50 mg/kg for spices and selected commodities; 2‐CE tolerances historically much higher compared with the EU	Same residue may be compliant in one market and noncompliant in another	Sharpens the international trade dimension of the EO problem
Codex Alimentarius	Consensus‐based international standard setting	Ongoing discussion of EO and 2‐CE without a fully harmonized resolution	Slow progress where residue definition and toxicology remain contested	Future harmonization depends on better occurrence, toxicology, and analytical comparability data

Abbreviations: 2‑CE, 2‑chloroethanol; Codex, Codex Alimentarius Commission; EFSA, European Food Safety Authority; EO, ethylene oxide; EU, European Union; MFDS, Ministry of Food and Drug Safety (Republic of Korea).

Source: Author‐compiled synthesis based on the literature cited in the text.

7. Research and Policy Priorities

7.1. Priorities for Toxicological Clarification

Future toxicological work should be designed with regulatory usability in mind. It is not sufficient to generate general data; evidence must address the specific decision points that currently divide authorities. These include whether 2‐CE can support a health‐based guidance value distinct from EO, how uncertainty factors should be justified in food‐related contexts, and how compound‐specific evidence should interact with precautionary defaults in enforcement settings. Studies that directly target these decision nodes are likely to contribute more to harmonization than additional evidence that remains difficult to map onto real‐world regulatory questions. Importantly, the study design should ensure that the route of exposure, endpoint selection, and uncertainty treatment are stated explicitly, enabling the resulting evidence to be translated more consistently into dietary risk assessment and enforcement practice.

The first priority is to clarify how 2‐CE should be treated toxicologically in dietary risk assessment. Current practice often relies on risk management assumptions that are understandable under precaution but do not fully resolve the underlying scientific question. More compound‐specific evidence on oral exposure, dose–response relationships, and appropriate health‐based reference values would strengthen the scientific basis of both national control measures and international standard‐setting discussions. This work is especially important because the present debate is no longer about EO alone, but about what can legitimately be inferred from marker compounds in food (O'Keeffe et al. ²⁰²²).

Such clarification should ideally address several linked problems: the adequacy of existing in vitro evidence for food‐related interpretation; the extent to which oral exposure scenarios differ from occupational or inhalation paradigms historically associated with EO; the relationship between structural analogy and regulatory equivalence; and the level of uncertainty that should be tolerated when direct compound‐specific evidence remains limited. A more explicit mapping of these issues would help authorities explain why a given interpretive stance has been adopted.

Importantly, additional toxicological work would have value even if it does not produce complete consensus. Better evidence can narrow the range of reasonable disagreement, improve transparency about uncertainty, and reduce the tendency for policy positions to be presented as if they were purely data driven when they also contain clear risk management judgments.

7.2. Priorities for Analytical Standardization

A broader view of standardization would also strengthen scientific publishing. Papers on EO methods are often strongest on instrumental detail and validation metrics, yet less explicit about how the resulting data should be interpreted in regulatory or supply chain contexts. Including a brief interpretive statement as part of method reporting—clarifying what the workflow is best suited to demonstrate and what it cannot reliably establish—would improve comparability across studies and reduce overinterpretation by downstream users. Because EO‐related results now routinely circulate beyond the laboratory into regulatory, commercial, and public‐facing domains, this kind of disciplined interpretive reporting should be regarded as part of analytical quality rather than as an optional discussion point. In practice, this would mean pairing validation metrics with concise statements on analyte identity, matrix scope, reporting basis, and foreseeable limits on causal interpretation.

The second priority is matrix‐specific analytical standardization. Laboratories require harmonized guidance on sample temperature control, matrix classification, analyte basis of reporting, and the circumstances under which EO‐equivalent calculations should be applied. Method documents already acknowledge the importance of temperature, extraction design, and matrix‐matched calibration, but these parameters should be embedded in reporting frameworks that make interpretation more transparent to regulators and supply chain stakeholders. Each result should explicitly state whether it reflects parent EO, 2‐CE, or a converted residue definition, and it should clearly identify the relevant matrix context (FSSAI ²⁰²³).

7.3. Priorities for Regulatory Communication and Harmonization

There is a practical need to distinguish communication intended for consumers from communication intended for regulators and industry. Consumer‐facing messages require simplicity, whereas technical decision‐making requires explicit discussion of analyte identity, matrix context, and uncertainty. If the same simplified message is used for both audiences, either technical precision is lost or public communication becomes opaque. EO incidents therefore illustrate a broader challenge in food safety governance: communication must be tiered to the audience while remaining internally consistent. This is especially critical when authorities rely on marker‐based evidence rather than direct observation of the original exposure event (European Commission ²⁰²¹). A tiered communication model would help preserve confidence while still allowing technical audiences to see where a result reflects direct measurement, conversion‐based reporting, or cautious inference from incomplete evidence.

Effective communication also depends on situating it correctly within the broader structure of risk analysis. Risk communication is not a downstream add‐on, but one of the three interdependent components of risk analysis, alongside risk assessment and risk management, and these components should be considered together. In the EO context, this means that communication should address both what the assessment shows—including the uncertainty attached to marker‐based inference—and what management actions follow and why, rather than compressing a complex evidentiary picture into a single reassurance or alarm.

Regulatory communication should make clearer which decisions are screening decisions and which are final risk management judgments. A screening result is designed to prompt further inquiry or temporary control; it does not necessarily establish a complete causal account. By contrast, a final risk management judgment carries broader implications for recall scope, importer accountability, and international trade. Explicitly distinguishing these stages would make EO control more transparent and proportionate. It would also create a clearer interface between national incident management and Codex‐level discussions about how residues should be conceptualized internationally (Codex Alimentarius Commission ^2024b; European Commission ²⁰²¹).

The third priority is to make source attribution an explicit part of EO incident management rather than an implicit assumption. Analytical findings should be interpreted alongside ingredient origin, processing conditions, storage history, and product architecture. This is particularly important for processed and composite foods, in which analytical soundness does not by itself establish the source event (Section 5.3). Regulatory communication should therefore differentiate among screening, confirmatory analysis, and causal interpretation (European Commission ²⁰²¹).

This recommendation has both technical and institutional dimensions. Technically, it requires better integration between laboratory reporting and supply chain investigation. Institutionally, it requires regulators to communicate uncertainty in a way that is transparent without undermining public confidence. These two demands are sometimes treated as conflicting; however, in EO management, they are better understood as complementary: transparent communication about what is known, what is inferred, and what remains uncertain is more likely to sustain long‐term credibility than a uniform yet overconfident narrative.

At the international level, harmonization efforts should focus less on forcing immediate numerical consensus and more on improving the comparability of evidence categories. Occurrence data, analyte definitions, reporting templates, toxicological assumptions, and matrix classifications all need to become more interoperable before regulatory convergence becomes realistic. In practical terms, a durable framework will require clearer separation among hazard identification, analytical marker interpretation, and risk management action.

The framework proposed in Section 5.3.2 (Tier 1–3 inference typology; Figure 4 and Table 3) provides a candidate near‐term objective. By aligning tier descriptors across jurisdictions, even where local risk management defaults differ, regulators could reduce the interpretive noise that currently complicates comparison. Such tier‐based alignment would not eliminate disagreement, but it would make disagreement more legible by clarifying which interpretive stage is actually contested.

FIGURE 4.

Source attribution framework for ethylene oxide‐related residue findings in foods. Each finding is evaluated along four analytical dimensions, combined into a tier classification and mapped onto a graded regulatory response: D1, analytical character (parent EO, 2‐CE, or both; basis of reporting); D2, matrix class (raw single‐ingredient, processed single‐ingredient, multi‐ingredient composite, or finished reformulated product); D3, supply chain and temporal context (origin documentation, supplier blending, elapsed time, and chain‐of‐custody); and D4, plausibility of alternative sources for the marker compound. The combined assessment (D5, inference strength) assigns the finding to one of three inference tiers: Tier 1 (strong), attributable to a specific upstream exposure event; Tier 2 (moderate), with established EO‐related chemistry but no identifiable event; and Tier 3 (weak), consistent with EO‐related chemistry but lacking event‐level support. The assigned tier informs a graded, proportionate regulatory response (D6). The framework supports transparent interpretive practice in incident management and complements jurisdiction‐specific risk management procedures. Source: Prepared by the authors based on the literature cited in the text.

7.4. From Incident Response to a Resilient System: Progress, Gaps, and Actionable Priorities

In the years since the 2020–2021 European EO incident, regulatory activity has shifted from emergency response toward routine control, and several concrete changes are now in place. The durability of this system, however, depends on closing a set of gaps that extend across the supply chain, from ingredient origin through processing and blending to the finished product. Framing the remaining work in supply chain terms clarifies not only what has improved but also where targeted action would yield the greatest benefit.

Upstream, the intensification of import controls on higher‐risk commodities, together with expanded lists of monitored crop–country combinations, has strengthened documented oversight for several product streams (European Commission ^2022a). Two gaps nonetheless persist. First, enforcement still depends heavily on documentation generated in exporting countries, and it does not yet distinguish systematically between residues arising from fumigation and the low‐level 2‐CE that can form through non‐fumigant routes (Section 3.3). Second, contamination has repeatedly entered finished products through blended ingredients and additives present at low inclusion levels, where traceability is weakest. Actionable measures include strengthening exporter qualification and origin documentation, applying risk‐based rather than uniform sampling, and extending ingredient‐level traceability and batch‐level certification to the high‐risk stabilizers and additives that have historically served as contamination vectors.

At the enforcement stage, harmonized crisis coordination procedures have made incident response more systematic, while routine multiannual monitoring now provides a clearer picture of where EO‐related residues occur. It should be recognized, however, that such monitoring is a form of passive surveillance rather than an active control measure: a commodity‐by‐year sampling schedule characterizes the scale and distribution of the problem but cannot, in itself, minimize contamination or serve as a fail‐safe barrier. What remains underdeveloped is the proportionality of that response. Regulatory action is not yet consistently calibrated to the strength of the underlying inference, and a trace 2‐CE finding near LOQ may in practice be interpreted as suggestive of fumigation. The tiered source attribution framework set out in Section 5.3 offers a practical basis for calibration: it links the evidentiary strength of a finding to a proportionate response, distinguishing screening‐level control from confirmed causal attribution and separating precautionary withdrawal from full recall. Embedding such a framework would make enforcement both more transparent and more defensible.

Two technical reforms would support this shift. Analytically, official methods should require each result to state its analyte identity, matrix class, and reporting basis—parent EO, 2‐CE, or a converted residue definition—so that findings become comparable across laboratories and jurisdictions (Section 7.2). Toxicologically, the pending EU‐level assessment of 2‐CE provides an opportunity to resolve the residue definition question directly. Establishing a health‐based guidance value for 2‐CE and adopting residue definitions that report EO and 2‐CE separately—as BfR has recommended and as US practice already reflects—would allow occurrence data to be interpreted against exposure‐based reference values rather than through hazard‐led detection alone (Sections 2.2 and 4.2; BfR ²⁰²⁶; EPA ²⁰²⁰).

Internationally, the most tractable near‐term goal is not immediate numerical consensus on maximum residue levels but improved interoperability of the evidence itself. Analyte definitions, reporting templates, matrix classifications, and toxicological assumptions all need to become more comparable before harmonized limits can be negotiated productively. Ongoing Codex discussions on EO and 2‐CE provide a natural forum for this work. Progress on comparability would reduce the trade friction that arises when the same analytical result is compliant in one jurisdiction and a violation in another (Section 2.2).

8. Conclusions

EO and 2‐CE in foods represent more than a residue monitoring issue. They highlight a broader challenge in modern food safety: the misalignment between what toxicology can establish with confidence, what analytical chemistry can detect with precision, and what regulatory systems must decide under conditions of uncertainty. As argued throughout this review, the EO issue persists not because evidence is absent but because different forms of evidence are often treated as though they answer the same question.

Three conclusions follow. First, EO itself remains a compound of clear toxicological concern, and strong exposure control is scientifically justified. Second, 2‐CE is an analytically important and policy‐relevant marker, but its role should be described more precisely to avoid conflating evidence of EO‐related chemistry with full toxicological equivalence to EO. Third, harmonization will remain limited unless food control systems explicitly distinguish among hazard characterization, residue measurement, source attribution, and matrix‐dependent regulatory interpretation.

The case of EO also illustrates a broader lesson for food safety science. As analytical methods become more sensitive and supply chains become more complex, controversies will increasingly center on how to interpret valid but incomplete evidence. In such settings, the credibility of food control depends not only on technical competence but also on conceptual clarity about what a result can and cannot prove.

Taken together, these findings suggest that the future of EO‐related governance will depend less on ever‐lower detection limits than on stronger interpretive frameworks. The tier‐of‐inference framework developed in Section 5.3.2, supported by the precedent comparison in Section 6.1, offers a candidate basis for shared interpretive language across jurisdictions and a structured approach to source attribution in processed and composite foods. Matrix‐specific analytics, clearer toxicological positioning of 2‐CE, and more transparent international decision rules will all be essential if EO‐related residue control is to remain both scientifically credible and practically feasible across increasingly complex food systems.

Author Contributions

Hyeonseo Choi: investigation, writing – original draft, conceptualization. Yujin Ahn: writing – original draft, data curation, investigation, visualization. Yooheon Park: conceptualization, funding acquisition, writing – review and editing, project administration, supervision.

Conflicts of Interest

The authors declare no conflicts of interest.

Data Availability Statement

No new data were generated or analyzed in support of this review.