Examples of Quantitative Mycotoxin Risk Assessments Use and Application in Risk Management


Mycotoxins are widespread chemical hazards in crops and plant products, which pose a risk to human and animal health. Although considered as chemical hazards, they have a biological origin, as they are produced by filamentous fungi. Their presence in our food may be of concern, and there is an urgent need for control strategies based on both qualitative and quantitative risk assessment.

A number of risk assessments, commonly based on dietary exposure point estimations, have been carried out for several mycotoxins, including aflatoxins, ochratoxin A and Fusarium mycotoxins, in different countries. The estimates have resulted, in general, in safe risk characterization situations despite the acknowledged limitations to accounting for related uncertainties and variabilities. In this regard, more refined approaches using stochastic exposure models have revealed that the exposure among several population groups, such as babies, infants or high consumers of certain food products, may be of concern. Several examples of the application of stochastic exposure and hazard assessment models will be discussed, including the advantages and drawbacks as well as the key methodological issues.

Quantitative risk assessment (QRA), based on food chain models instead of analytical data from food samples, although hardly explored for mycotoxins, allows the estimation of different simulated scenarios of exposure throughout the food chain. This includes cropping, postharvest and food processing. In this way, it is possible to test different risk management alternatives, run the simulations and choose the alternative that results in a lower level of exposure for the population. As mycotoxins have a biological origin, the models used for the exposure assessment may include predictive mycology models based on fungal growth and mycotoxin production.


Mycotoxins are secondary metabolites produced by several fungal species. The worldwide contamination of foods and feeds with mycotoxins is a significant problem. From an economic and public health standpoint, the foodborne mycotoxins that are considered as being relevant are afla- toxins, fumonisins, certain trichothecene mycotoxins (including deoxyni- valenol [DON] and T-2 and HT-2 toxins), ochratoxin A (OTA), patulin and zearalenone (ZEA).

Mycotoxins can cause a variety of adverse health effects in humans. Aflatoxins, including aflatoxin B1 (AFB1), are the most toxic and have been shown to be genotoxic; i.e., they can damage DNA and cause cancer in animal species. There is also evidence that they can cause liver cancer in humans. Other mycotoxins have a range of other health effects, including kidney damage, gastrointestinal disturbances, reproductive disorders and suppression of the immune system. For most mycotoxins, a tolerable daily intake (TDI) has been established, which estimates the quantity of mycotoxin to which someone can be exposed daily over a lifetime without it posing a significant risk to health.


Risk assessment is commonly structured as a conceptual framework that in the context of food chemical safety, provides a mechanism for the systematic compilation, integration and critical review of information relevant to estimating the probability of adverse health effects derived from exposure to harmful chemicals present in food. For operational reasons, the most widely implemented risk assessment frameworks for mycotoxins divide the process into four major components: hazard identification, hazard characterization (or dose-response assessment), exposure assessment and risk characterization.

The extent of precision of the analysis relies on the quality of primary data on the toxicity of the compound (hazard characterization) and concerning the population exposure (exposure assessment); hence, the overall process can include a key component in which the probability of harm is estimated. In general, the risk assessment frameworks follow a tiered approach: qualitative assessments (Tier 1), deterministic assessments (Tier 2) and finally, probabilistic assessments (Tier 3) (EFSA 2008). In this sense, the probabilistic assessments, also known as stochastic, refer to the quantitative analysis of variability and uncertainties related to the estimated health risks of chemicals, depicting the QRA approaches (Figure 17.1).

The QRA frameworks have been commonly developed under regulatory settings pursuing the achievement of health protection goals extending or complementing the deterministic or point estimate risk assessment

Representation of the variability of exposure estimates and the uncertainty distributions for given percentiles

Figure 17.1 Representation of the variability of exposure estimates and the uncertainty distributions for given percentiles (e.g., 75th and 95th) estimated through Monte-Carlo Simulations. Adapted from Council et al. (2005).

frameworks. Overall, public policies developed for managing food chemical risks are generally precautionary rather than being focused on producing accurate predictions. Hence, the default approaches are mainly deterministic, based on "worst-case" scenarios, using conservative uncertainty factors, resulting in margins of safety that do not require more refined (probabilistic) evaluations. In consequence, QRA of food chemicals and specifically of mycotoxins has been mainly conducted within minor academic or scientific contexts rather than being promoted by food protection agencies. Actually, there is an acknowledged lack of standardized guidelines or frameworks for the QRA of mycotoxins delivered by scientific supporting agencies like the European Food Safety Authority (EFSA). Only one Scientific Opinion from EFSA provides insights into the application of probabilistic methods for modelling dietary exposure to pesticide residues, which may partially apply to mycotoxins (EFSA, 2012).

From a theoretical point of view, the QRA framework can be structured on the basis of the deterministic workflow, whereby a probabilistic assessment may be performed on the exposure distribution parameters and also extended to the dose-response assessment (hazard characterization), and both may be integrated into a full probabilistic risk assessment setting (Figure 17.2). Since no definite settings have been established, QRA of mycotoxins has been mainly focused on the probabilistic exposure assessment and in minor cases, on probabilistic hazard characterization.

Dietary Exposure Assessment of Mycotoxins

The absence of fully validated individual biomarkers for mycotoxins constrains the exposure assessments to dietary exposure modelling by integrating consumption with occurrence data.

In general, dietary exposure assessment describes the pathways through which mycotoxins are introduced into the food chain and challenged throughout the food processing and production systems, their distribution within the food commodities and finally, their consumption by the end-users. However, different objectives may motivate an exposure assessment, and consequently, the approach and methodology may vary. The aims may be, for example, to be combined with a hazard characterization as part of a risk assessment to estimate the risk associated with a mycotoxin+commodity combination, to identify foods in the diet likely to make a major contribution to human exposure to hazards, to identify where interventions or control options are likely to be most effective in reducing the level of exposure to a hazard in a given product, to compare

Schematic representation of a full probabilistic risk assessment framework for mycotoxins

Figure 17.2 Schematic representation of a full probabilistic risk assessment framework for mycotoxins.

the efficiency of mitigation measures in reducing the exposure to a given hazard or to compare the levels of exposure resulting from different processes and food products (WHO, 2008).

Thus, depending on the scope, exposure assessment can begin with the evaluation of mycotoxin contamination at field level or in raw materials (e.g., a "farm-to-fork" risk assessment), or it can begin with the description of the mycotoxin contamination distributions at subsequent steps (e.g., as input to a food processing step) or just at the point of consumption. In any case, the main goal of risk assessment is to estimate the likelihood of mycotoxins being ingested by the consumer. By completing the pathway to the consumer and the potential adverse health effects, we incorporate the important information that may be critical for depicting the food management strategies along the food chain (i.e., modification of storage conditions to impair the fungal growth or mycotoxin production).

As introduced earlier, most of the existing dietary exposure assessments of mycotoxins have consisted of combining food consumption distributions deterministically or probabilistically with the occurrence of such mycotoxins in those target food categories; thus, they assessed the exposure of the population under the current risk management situation. It must be highlighted that exposure assessment can be conducted more accurately when mycotoxin concentrations are measured in the consumed product instead of predicted by a model. On the other hand, little research has considered the whole food chain or particular processing steps to link exposure assessment to risk management, due possibly to the lack of suitable data both on models for the prediction of mycotoxin production and on the impact of processes in mycotoxins to build risk assessment models.


Conceptually, the exposure to mycotoxins may be better described as a range of values rather than a single value (i.e., a deterministic approach), because each individual in the population is expected to be exposed to different concentrations of mycotoxins over time, displaying a group of random variables that may depict the classical definition of a stochastic process.

The general exposure model used to assess intakes from the diet can be written as in Equation 17.1:

where i is an index for an individual (i = 1E( is the usual intake of individual i (ng/kg body weight/day); t is an index for the time window to assess food consumption and contamination (weeks [W], days [D] or single occasions [O] of consumption; g is an index for food group (g = 1,...«); Qi,s is the consumption of food group g on occasion t by individual i (kg); CiXg is the contamination of food group g encountered on occasion t by individual i (pg/kg); n, is the number of days of food records available for individual i; and bio, is the body weight of individual i (kg).

In the probabilistic exposure model, consumption (Q; ( J) and contamination levels (C, ,J in Equation 17.1 are drawn randomly from underlying cumulative distribution functions (CDFs) in a one-dimensional Monte- Carlo simulation through two main approaches:

  • • Non-parametric method (NP). In the NP approach, no probability density function (pdf) hypothesis is made either on the consumption or on the contamination data. Each normalized consumption profile of the survey is taken into account, and each type of consumed food is assigned a value of contamination randomly drawn from the available contamination data.
  • • Parametric (P) and mixed parametric methods (MP). The P approaches attempt to better fit the pdf to the contamination and consumption datasets. In the MP method, a mixed pdf is fitted to each food consumption and a parametric pdf to each food contamination dataset.

Common approaches to evaluate the best fit of candidate distributions include the maximum likelihood method, quantile-quantile (Q-Q) plots, the Anderson-Darling statistic and the agreement between empirical and estimated quantiles.

< Prev   CONTENTS   Source   Next >