Conclusions from the workshops on Ochratoxin A in Food: Recent developments and significance, organized by ILSI Europe in Baden (Austria), 29 June–1 July 2005

In 1996, the International Life Sciences Institute (ILSI) organized the first workshop on “Ochratoxin A in Food” in Aix-en Provence, France, with the aim to discuss the occurrence of this mycotoxin in different food commodities, its fate during processing, analytical methods for detection, human exposure and the toxicological effects associated with exposure to ochratoxins. The aim of the present workshop in the early summer of 2005 was to review new data that have emerged since the previous meeting with an international forum of experts. Special attention was given to the results of the EC 5th RTD Framework Program funded projects that were devoted to “Prevention of OTA in cereals and mechanisms of OTA-induced carcinogenicity as a basis for improved risk assessment”, and to the projects on “Amelioration of contaminations” initiated by the FAO.

The workshop comprised a series of lectures presenting data on occurrence and prevention of toxin contamination, sampling and analytical techniques for toxin quantification, assessment of human exposure, epidemiology of diseases that have associated with toxin exposure, and new toxicological data and their possible impact on risk assessment. During the workshop, the participants were allocated to three different working groups given the task to discuss and summarize the new data and to present their conclusions on the following aspects:

• Appropriateness of current sampling and analytical methods to assess the degree and level of food contamination;

• Assessment of the current knowledge on human exposure to OTA and potential biomarkers of exposure; and

• Assessment of the epidemiological data on disease conditions that have been associated with OTA exposure such as Balkan Endemic Nephropathy (BEN) and Urinary Tract Tumours (UTT);

• Evaluation of the recent data on the mechanisms associated with OTA toxicity and carcinogenicity.

A final plenary discussion at the end of the workshop reflected the data presented by the invited speakers and the discussions of the delegates during the meeting. These discussions provided the basis for this brief summary.

Occurrence, analytical control measurements and methods of prevention

Previously, it was assumed that OTA is predominantly formed by P. verrucosum in temperate climates, and by A. ochraceus in warmer regions. More recently, the important role of Aspergilli of the section Nigri, and especially A. carbonarius has been recognized as key species responsible for the contamination of grapes, wine and vine fruits with OTA. A. carbonarius may invade grapes already at the pre-harvest stage, particularly when the production of vine fruits (Corinth's currants) includes a field drying process. Therefore, efforts to minimize OTA contamination have to start with the effective management of this drying process. However, the observed pre-harvest contamination also implies that toxin formation cannot be entirely avoided, as the rate of contamination depends on the climatic conditions at the pre-harvest stage. Correct management of harvested materials, additional drying and cleaning procedures, as well as control measurements to assess the prevalence of toxin formation, need to be implemented. Logistical models to define the accurate conditions for storage and processing (a_w and temperature) have been already developed for other food commodities, particularly grains. Recent data confirm that thorough cleaning of grains and milling reduces the OTA concentrations in end products. The by-products from these processes have been used as animal feeds in the past, and hence have to be considered as source of animal exposure.

It is generally agreed upon, and reinforced by information collected with the EU SCOOP tasks (EC, 1997 SCOOP task 3.2.2 and EC.2002, SCOOP task 3.2.7), that cereals are the most prominent source of OTA in the human diet. Other recognized sources are coffee and wine, but it needs to be recalled that less prominent sources, such as spices, beer, cocoa, dried fruits (for example) figs, liquorice and grape juice as well as pig (blood-based) products and occasionally poultry meats may contribute to human exposure, particularly in certain high-consumers subpopulations. Hence analytical methods as well as representative consumption data for these commodities need to be established and evaluated to allow monitoring of commercial food commodities.

Discussing the applied analytical methods to control the occurrence of OTA in foods, it was agreed that in principle accurate and reliable methods are available. Based on literature data and on the FAPAS® proficiency testing programme, it can be stated that HPLC with fluorescence detection (FD) coupled with immuno affinity (IA) column or solid phase extraction (SPE) clean-up, is by far the most widely used technique. Harmonized sampling policies for surveillance programs should follow.

Statistically-based sampling plans take into account the distribution of OTA in a given commodity, geographically-based risk factors, and the overall contribution of a given food product to human exposure. Targeted sampling can easily be implemented by industry (and controlled by the respective authorities), but the techniques used need to comply with the expected (in-) homogeneity of the distribution of the toxin in the food commodity under consideration. In the final discussion it was stated that both, sampling techniques and analytical methods, should be evaluated for their fitness-for-purpose prior to implementation.

In summary, the ad-hoc working group concluded that:

• While most EU member states reported data on the occurrence of OTA in breakfast cereals, in cereal-based food for babies and infants, coffee, beer and wines (see SCOOP task), there is a lack of information on the OTA contamination of certain minor food commodities (including products intended for children's consumption) such as cocoa-derived products, liquorice, grape juices, dried fruits (figs, raisins and currants) as well as spices, and teas;

• Presented datasets should clearly indicate the nature and purposes of analytical controls, i.e. targeted sampling (based on identified risk factors) or at random sampling (exposure and quality assessment). Hence harmonized guidelines for statistic-based sampling strategies and the uniform reporting of analytical results should be implemented;

• Prevention strategies (including HACCP approaches covering safe production, harvesting, storage, transport and processing) need to be developed and implemented for all food commodities at risk for OTA contamination;

• OTA is partially degraded in some commercial processes. The nature and toxicity of the degradation products remains to be elucidated;

• A number of validated methods are available for OTA analysis in a variety of food matrices. However, there is still a need for validation of rapid methods that allow real-time analysis at food production sites. Certified reference materials at levels close to OTA regulatory limits are needed for all relevant foods.

Exposure assessment and biomarkers of exposure

Surveillance on the occurrence and prevalence of OTA in food commodities forms the basis for exposure assessment and contributes to hazard characterization. Using the data set from the French monitoring system for OTA contamination of raw commodities in which samples with non-detected levels are assumed to be half of the LOD and the food consumption data from France (INCA database – Enquête Individuelle et Nationale sur les Comsommations Alimentaires 1999), different probabilistic scenarios could be compared. Results show that the distribution of exposure predominantly depends on the efforts put into converting foods at the raw-edible-level (to which most of the analytical data from surveillance programs refer) into foods-as-consumed for which individual information about quantities is regarded. Applying a Total Diet Scenario (TDS) the (over-)estimation of exposure is corrected in two ways, by considering the concentration of OTA in food-as-consumed, and by using data from pooled samples correcting for the fact that a consumer is unlikely to consume highly contaminated food regularly. Hence, it was agreed during the workshop that the use of extended recipes and/or total diet studies are the preferred methods resulting in a more realistic approach of human exposure assessment.

Human dietary exposure has been confirmed by measuring OTA plasma/serum levels in different countries and distinct segments of the population (rural versus urban regions). In some occasions urine levels have been measured also. In answering the questions, which are the most reliable biomarkers of exposure that could be associated with evidence of disease conditions attributed to OTA, the ad-hoc working group concluded that:

• OTA levels in plasma and urine have to be considered as typical biomarkers of exposure. No other appropriate effect-related biomarkers have been identified as yet. Determination of OTA-DNA-adducts in peripheral lymphocytes and detailed analysis of exfoliated cells in urine samples of human patients could be considered as biomarkers indicative for toxic effects;

• Exposure monitoring should be further standardized with respect to timing of sampling, number of repeats and analytical methods. A prerequisite for advanced sampling is the better understanding of the kinetic parameters characterizing the disposition of OTA in human tissues;

• Exposure assessment should be based preferentially on probabilistic modelling for which an adequate database is a prerequisite. Whenever possible, exposure assessment should address particularly vulnerable populations, such as the pregnant and breast-feeding mothers, the neonate, children in general (having often specific consumption patterns) and human patients with impaired renal function.

Human epidemiological data

The final task of hazard characterization is the assessment of the possible link between the consumption of food contaminated by OTA and human diseases. As yet, two distinct pathological conditions have been associated with exposure to OTA: BEN (Balkan Endemic Nephropathy) described as progressive karyomegalic interstitial nephritis resulting ultimately in complete renal failure, and UTT (Urinary Tract Tumours) that have been reported to occur with a higher incidence in endemic areas of BEN. Previous reviews of the available data on the toxicology of OTA have stated already that in all mammalian species tested, the principle target of OTA toxicity is the renal proximal tubule. Renal carcinogenicity has been clearly demonstrated in rodent species at doses exceeding those that induce nephrotoxicity. This observation together with the LOEL for nephrotoxicity in pigs (found to be the most sensitive animal species) was used for safety evaluation and the establishment of Provisional Tolerable Daily Intake varying between 1.2–5.7 ng/kg b.w. (WHO 2001) or recommended to be <5 ng/kg b.w. (SCF 1996, 1998). Previously, the International Agency for Research on Cancer (IARC 1993) had classified OTA as possible human carcinogen (group 2B), based on the sufficient evidence for carcinogenicity in animal studies, but inadequate evidence in humans.

In a reappraisal of the evidence for human carcinogenicity associated with OTA exposure, the working group concluded that in the recent years no additional data have become available. Hence, there is still insufficient evidence in support of carcinogenic effects of OTA in humans. It was strongly recommended to make tumour material, which obviously has been collected in the endemic areas of BEN, available to the scientific community. Classification with advanced methods of the tumours collected from endemic areas would allow extrapolation from animal data to humans. Moreover, detailed investigations of kidney specimen from BEN and UTT patients may also indicate or exclude the association of these diseases with other nephrotoxic compounds, including medicinal products such as non-steroidal anti-inflammatory drugs (NSAIDs) or certain nephrotoxic plant-derived remedies.

Reviewing the evidence from human epidemiology that is indicative for an association between exposure to OTA and the prevalence of associated renal diseases, the ad-hoc working group concluded that:

• There is no convincing evidence from human epidemiology to confirm the association between OTA exposure and the prevalence of BEN or UTT. The existing epidemiological data are not sufficient to drive risk assessment;

• Human primary kidney cells are highly susceptible to OTA toxicity and data from both, in vitro as well as in vivo experiments, make it plausible that OTA comprises a risk to human health. In pigs (considered to be the most sensitive animal species) a direct correlation between OTA exposure and onset and progression of nephropathy (loss of parenchymal cells and replacement with mesenchymal tissue) have been found. These findings are in line with the above-mentioned data from in vitro experiments with human (and porcine) cells;

• There is a need for a re-evaluation of kinetic data to improve the understanding of species and gender differences and the dynamic doses at the site of action, considering that target cell populations might change their characteristics during exposure (de-differentiation). Distinctions should be made between intermittent peak exposures versus chronic low dose exposure. Moreover, exposure to OTA at early stages of life deserves attention, as the developing organism might be more sensitive than the adult with respect to onset of disease;

• Regarding UTT, there is a need for a better characterization of the tumours observed in the endemic areas with respect to their renal cell or transitional cell origin. Detailed investigation of tumour material would allow for a comparison with the findings from the carcinogenicity studies conducted in rodents;

• The potential role of NSAIDs and other compounds exerting proximal tubule damage in the aetiology of BEN and UTT needs to be clarified.

Mechanisms associated with the carcinogenicity of OTA: Evidence for genetic and epigenetic mechanisms

Since the latest evaluations by JECFA (2001) and SCF (1998) new data have become available on the mechanisms of toxicity associated with OTA exposure. In particular, the results obtained with the 5th RTD Program (QLK1-2001-011614) where presented and discussed during the workshop.

New data indicated that neither OTA nor OTB (given as ¹⁴C-labelled toxins to male F344 rats at single dose of 500 µg/kg b.w.) resulted in the formation of covalent DNA adducts detectable by ¹⁴C-AMS (Accelerator mass spectrometry: LOD < 3 adducts/10⁹ nucleotides for OTA, and <10 adducts/10⁹ nucleotides for OTB). In addition, in experiments in which high doses of OTA (up to 2 mg/kg b.w.) were given repeatedly for a period of two weeks to male F344 rats, LC-MS/MS analysis failed to detect the previously postulated OTA-dG adducts, although synthetic OTA-dG and OTA-dGMP adducts were available as reference. In the same experiments, however, DNA-strand breaks were found in the liver, kidney and spleen of animals treated with OTA, and interestingly, also in animals treated with OTB.

In contrast, using ³²P-post labelling technique, OTA was found in different in vivo and in vitro experiments to induce DNA lesions. Various degrees of evidence suggest that these DNA lesions are linked to the expression of biotransformation enzymes including CYP1A2, CYP2C9, CYP3A4 and probably CYP1B1, and/or to other oxygenases, such as cyclooxygenase (COX) and lipoxygenases (LOX). Consistent with these findings, DNA-lesions as detected with ³²P-post labelling were found not only to be time- and dose-dependent, but also to be modulated by inducers and inhibitors of these enzymes. In an attempt to identify the chemical nature of the ³²P-labelled DNA lesions it could be shown that individual renal DNA-adducts detected in OTA-exposed rats co-migrate on TLC plates with synthetic (deoxy)guanosine OTA-DNA adducts (O-C8-dG-OTA and C-C8-dG-OTA).

New data were also presented supporting the hypothesis that OTA carcinogenicity is related to distinct epigenetic mechanisms. The formation of reactive oxygen species as induced by OTA had been previously described and associated with oxidative DNA damage. The presented new data are supporting this hypothesis by providing evidence that OTA is affecting several cell signalling pathways including the activation of mitogen-activated protein (MAP) kinases, extracellular signal regulated (ERK 1/2) kinases and C-jun amino terminal (JNK 1/2) kinases in renal cells in vitro. Moreover, new toxicogenetic data, based on assessment of the gene expression profiles in rats that were exposed for a period of up to two years to OTA (at concentrations equivalent to approximately 300 µg/kg b.w.) indicated that OTA exposure alters the expression of genes associated with cellular calcium homeostasis and interferes with pathways regulated by the transcription factors HNF4α (hepatocyte nuclear factor 4 alpha) and Nrf2 (Nuclear factor erythroid 2-related factor 2). Previous data had suggested already that an impairment of HNF4α pathways can be associated with renal carcinogenicity, whereas the depletion of Nrf2-regulated enzymes is associated with an impairment of the defence potential of a cell against oxidative stress. Taken together, these data provide suggestive evidence for a role of oxidative stress in renal pathologies associated with OTA, including the presence of epigenetic mechanisms in the development of renal tumours.

The ad-hoc working group summarized the current findings on genetic and epigenetic mechanism as follows:

1. Renal-specific toxicity of OTA and evidence in support of epigenetic mechanisms

   • OTA is able to induce renal cell necrosis, apoptosis, karyomegaly and cell hyperplasia, and increases mitosis;

   • OTA inhibits protein synthesis, produces oxidative stress, induces mitochondrial dysfunction, and modulates cell signalling and cell–cell interactions in a dose/concentration dependent manner;

   • The biological effective tissue concentrations of OTA depend on the overall oral bioavailability and on cellular uptake and excretion mechanisms. These mechanisms determine the tissue specific accumulation and hence the ultimate intracellular toxin concentration.

2. Evidence in support of genotoxic mechanisms

   • Convincing evidence indicates that exposure to OTA results in DNA damage in kidney, liver, testis, spleen, as well as lymphocytes, thymocytes and fibroblasts of various species;

   • DNA damaging effects have been demonstrated by means of the comet assay, UDS, micronucleus assay, abasic sites and 8-oxy-dG adducts as well by the ³²P-postlabelling technique;

   • Although there is evidence for a time- and dose-dependent induction of DNA-lesions in vivo when applying the ³²P-postlabelling technique, OTA-adducts could not be detected by AMS or liquid scintillation counting at a comparable level of sensitivity;

   • The chemical identities of adducts as detected by ³²P-postlabelling need to be elucidated, but it has been demonstrated that two of these adducts co-migrate on TLC plates with synthetic O-C8-dG-OTA and C-C8-dG-OTA. The intensity and profile of the described DNA adducts varied in different biological systems and could be modulated.

In conclusion, considering the consistent human exposure, including possible pre-natal (OTA crosses the placenta barrier in animal models) peri-natal (OTA has found to be excreted in breast milk) and nutritional exposure in later stages of life, attempts to characterize the risk associated with OTA exposure rely on the understanding of the mechanisms involved in OTA renal toxicity and carcinogenicity, as epidemiological data are too incomplete to drive risk assessment. The findings presented at the recent ILSI Workshop provide an updated view of the current insights regarding exposure assessment and risk characterization for OTA. At the same time, gaps in the current knowledge were identified, particularly regarding the lack of comprehensive toxicokinetic (PB/PK) data to assess human tissue concentrations, detailed and advanced descriptions and the classification of human renal pathologies commonly associated with exposure to OTA, and the molecular mechanisms underlying these diseases.