Current issues and perspectives in food safety and risk assessment

Abstract

In this review, current issues and opportunities in food safety assessment are discussed. Food safety is considered an essential element inherent in global food security. Hazard characterization is pivotal within the continuum of risk assessment, but it may be conceived only within a very limited frame as a true alternative to risk assessment. Elucidation of the mode of action underlying a given hazard is vital to create a plausible basis for human toxicology evaluation. Risk assessment, to convey meaningful risk communication, must be based on appropriate and reliable consideration of both exposure and mode of action. New perspectives, provided by monitoring human exogenous and endogenous exposure biomarkers, are considered of great promise to support classical risk extrapolation from animal toxicology.

Preface

Since ancient times food has played a pivotal role in the development of human society. Food security, synonymous for a reliable and adequate supply of healthy and safe foods, not only means freedom from hunger. By relieving individuals, families and populations from the constraints of laborious daily food procurement, it can be seen as a main catalyst, allowing for individual and societal development and education. Another beneficial aspect relates to the continuing technologic evolution inherent in modern food production, processing, storage, and distribution. This provides the consumer with an ever increasing multiplicity of high-quality and innovative food items. This freedom of choice, something never previously achieved, allows for individualized nutrition and brings about additional hedonistic value. Thus, food security in fact may be seen as a prerequisite for the development of society.

Food safety is an inherent element of food security. However, the latter term more specifically means that food as normally consumed should not pose health risks to the consumer. On a global scale, inadequate techniques of food production, storage, processing, and distribution today pose substantial risks to both food security and food safety. Food security is still at risk in underdeveloped regions of the world and starvation is far from being eradicated even though in theory conventional food production should suffice to feed today’s world. On the other side, inadequate distribution of resources governing societal food supply has led to major detrimental public health effects. Adiposity-associated diseases as a consequence of overnutrition of large parts of populations in developed (Western) countries today represent major public health risks, unrelated to food safety.

Hazard versus risk-based approaches

The basic methodology of safety evaluation on the basis of animal experiments has changed little during the past decades. Repeat dose toxicity is studied in laboratory animals, in compliance with established guidelines, mostly within the frame of 90-day subchronic or 2-year chronic toxicity studies. They inform us about a dose level associated with the non-observed adverse effect level (NOAEL).

By application of canonical uncertainty (safety) factors (mostly 100), an acceptable/tolerable intake level is derived from the NOAEL. The response in experimental animals is evaluated by using general, clinical chemistry, hematological, and histological parameters as indicators of organ damage and further parameters when applicable. The continuing progress in genomics, transcriptomics, proteomics, and metabolomics in combination with novel tools in bioinformatics and system biology has brought about promising new avenues toward improved toxic hazards characterization. Very often, the decision for in vivo testing is based on previously obtained in vitro evidence that calls for in vivo substantiation. However, in vitro systems provide much more information, especially with respect to mechanistic elucidation of toxicity outcomes. In this respect, the pivotal importance of in vitro systems for hazard characterization cannot be overemphasized. Complementary to in vivo toxicity testing, appropriate in vitro toxicological systems have eminent value and not only provide enhanced predictivity for hazards posed by food-associated chemicals. ¹ They are also uniquely important for characterizing key (bio)molecular events governing toxic outcomes in animals and/or humans. Such in vitro mechanistic studies are also of great value when there is suspicion or a structural alert with respect to genotoxicity.

In case there is a genotoxic carcinogen present in food, the European Food Safety Agency (EFSA) recommends implementation of appropriate mitigation measures to minimize consumers exposure down to a level in food resulting in a margin of exposure (MOE) of not less than 10,000. This MOE is deemed of low concern with respect to human health. ² The MOE is calculated by dividing the benchmark dose (BMD) derived from an appropriate carcinogenicity study associated with a benchmark response, e.g. 10% tumor incidence at the lower bound confidence interval (BMDL10), through the estimated human exposure level.

Thus, safety assessment of food constituents and/or contaminants not only requires the evaluation of a hazard potentially exerted by a specific compound but also needs to take into account the exposure of the consumer.

By definition, a hazard in food means a biological, chemical, or physical agent present in food that may have an adverse health effect. The term also encompasses an inherent property of an agent or situation having the potential to cause adverse effects. By contrast, the term risk describes the probability of an adverse effect and its magnitude in an exposed system or (sub)population.

At present, legislation in the European Union and elsewhere often includes a mix of both hazard-based and risk-based approaches for safety assessment. Under specific circumstances, for example, when immediate estimates of potential health concern are required, a hazard-based approach can be of value. This may for instance apply to emergency situations created by cases of accidental contamination or food adulteration, especially when human exposure is estimated to be close to thresholds for adverse effects. Sporadic examples may be mentioned here, for example, historic examples of adulteration of wine/spirits with diethylene glycol/methanol, adulteration of milk with melamines or, currently coming into focus, the illegal presence of drugs in food supplements. ³

Hazard-based approaches also may apply to agents exerting potent non-thresholded effects, such as certain strong genotoxic carcinogens. Of note, a hazard-based element also is intrinsic to the so-called threshold of toxicological concern (TTC) concept. This concept has evolved over the years, in regulatory authorities and elsewhere, as an attempt to develop a generic approach to the safety assessment of groups or individual chemicals of unknown toxicity, that is, with no or insufficient toxicological data. ⁴ The concept provides guidance about achieving acceptable risks by deriving toxicologically insignificant exposures according to hazard and chemical structure. ^{5 –7} This approach was introduced by the US Food and Drug Administration as threshold of regulation (1995) for compounds migrating into food from packaging materials. ⁸ It is based on structure-guided grouping of the compounds in question with respect to their expected toxicological properties in comparison with well-defined and studied reference compounds (‘Read across’). In brief, on the basis of grouping well-investigated compounds into different structural classes with respect to no-observed toxicological effect dose levels, three so-called Cramer classes (I, II, and III) ⁹ were generated. Toxicologically insignificant exposure levels were derived using the 5th percentile of each class, divided by a conventional uncertainty factor of 100. For hazards in foods associated with chemicals bearing structural alerts of genotoxic activity, the toxicologically insignificant exposure level was set at 0.15 µg/kg.

Zero tolerance of risks is not feasible for the majority of foods and their constituents/contaminants. In general, therefore, for food safety assessment, risk-based approaches are adequate. Such risk assessment of food has to consider man-made chemicals, as well as naturally occurring compounds, which might influence human health. The first prerequisite for risk-based approaches is a reliable exposure estimate, taking into proper consideration uncertainties in exposure assessment. The second important prerequisite comprises elucidation and appropriate consideration of the mode of action (MOA). Importantly, the MOA is required to be put into perspective with an appropriate estimate of consumers’ exposure. Furthermore, there is need to establish a continuous dialogue with the consumer in order to ensure appropriate risk communication, thereby creating societal confidence in the risk assessment. Moreover, any risk assessment appropriate for society should also take account of the benefits, including potential consequences for food security, sustainability, nutritional value, and others. The aim of such risk–benefit analysis should be to arrive at a science-based societal consensus concerning the acceptability of a given risk. This is quite important, given that consumers usually are risk averse, especially about exposure to chemicals in food.

Exposure assessment

Reliable metrics relating to the amount of an agent present in food; its bioavailability from the food matrix; and the duration, magnitude, and frequency of exposures are major determinants of the knowledge needed for risk assessment. ¹ Concerning food additives and ingredients, there may be substantial variation in formulations from standard recipes taken as basis for daily average dietary exposure estimates. This is a shortcoming recognized by most regulatory agencies, and it is also of relevance for exposure assessment of contaminants or residues. There is therefore a need for sustained development of methodology, encompassing aggregate exposure from all routes.

Exposure assessment is undergoing a fundamental change, as new tools become available, based on methodology of extreme specificity and sensitivity. Significant progress has been made toward monitoring biomarkers of exposure in humans. Presently, we witness the advent of metabolomics, techniques enabling us to picture the totality of metabolite spectra in a given body fluid or compartment, thus opening the possibility of collecting comprehensive analytical information about specific food intakes and their interaction with the organism. ¹⁰

As yet however, the monitoring of selected biomarkers is the preferred option. It requires detailed knowledge of the metabolism of the compound in focus to be able to use specific metabolites in body fluids or tissues as quantitative exposure indicators. For human studies, metabolite monitoring in urine and blood appears to be the method of choice.

An example is provided by the considerable database already available of human exposure estimates for the genotoxic carcinogen acrylamide (AA), based in part on biomarkers of exposure. ¹¹ AA is generated during heat treatment of food from innocuous precursors, namely, asparagine and a reducing sugar during the course of the Maillard reaction. AA itself is not a genotoxic agent but is converted in the organism by CYP450 to the corresponding genotoxic epoxide, 2,3-epoxypropane amide (glycidamide, GA). The latter reacts with DNA, forming DNA base adducts, predominantly at N7 of guanine. AA and GA react covalently with a spectrum of nucleophilic centers in biopolymers and plasma constituents and to a major extent also with glutathione. The glutathione adducts are terminally excreted in the urine as the corresponding mercapturic acids. ^{12 –14} Thus, the main biomarkers for internal exposure to the parent compound and its genotoxic metabolite, GA are:

urinary mercapturic acids (MAs),

hemoglobin (Hb) adducts of AA and GA, and

DNA adducts of GA.

Of note, monitoring these biomarkers not only allows for aggregate dietary exposure estimates. The direct comparison of MA metabolites and/or Hb adducts also indicates relative rates of toxification and detoxification at diet-related, low-exposure levels.

Uncertainties associated with assessment of consumer exposure to AA were recently adressed in detail in an European Food Safety opinion. ¹¹

It was recommended that:

reporting of AA occurrence data regarding the mode of preparation the products before analysis be improved;

duplicate diet studies be undertaken to provide more accurate indication of AA levels in food as prepared and consumed; and

data on urinary metabolite levels from individuals participating in duplicate diet studies be generated for the purpose of biomarker validation.

Beyond AA, these recommendations paradigmatically identify future research needs for the development and validation of biomarkers as adequate metrics for aggregate consumers exposure to genotoxic agents.

DNA adducts induced by genotoxic agents today are amenable to specific quantitation at levels of around 1 adducted DNA base in 10⁻⁸ DNA bases. This allows direct measurement of DNA damage (e.g. in peripheral white blood cells) associated with dietary intake of a genotoxic carcinogen at very low levels of food contamination (low µg/kg range). New methods increasingly become available that allow specific simultaneous detection of multiple DNA adducts in human tissue samples, including those formed by endogenous lipid peroxidation. ¹⁵ A further development, baptized “adductomics,” includes DNA adducts of known and of unknown identity. ^16,17

This rapid technological progress offers new perspectives for assessment of the risk associated with dietary exposure to genotoxic contaminants at levels of present-day consumers dietary intake. The induction of DNA damage can be viewed as the first step in the sequence of cellular key events leading to fixation of a mutation and eventually to malignant transformation. However, it has to be kept in mind that not every DNA lesion translates into a mutation since a spectrum of efficient repair systems are in place before cell division is achieved and a mutation fixed. Therefore, dose–response relationships of DNA damage induction and of mutation induction are not the same. It may thus be inferred that monitoring of DNA adducts as an exposure biomarker is a conservative approach with respect to potential biological consequences. Yet it provides a very significant step forward since reliable dosimetry at low levels of consumer exposure can now be achieved, obviating the necessity for high to low (“virtually safe”) dose extrapolation mainly used until now in classical risk assessment of genotoxic agents.

Another aspect largely disregarded until now comes from our increasing knowledge concerning the substantial background of DNA damage in human tissues, consistently induced by endogenous genotoxic agents generated during normal energy/housekeeping metabolism. It has been shown that a number of such background DNA adducts are identical to those arising in humans by exposure to exogenous genotoxic agents, including, for instance, ethylene oxide, vinyl chloride, formaldehyde, acetaldehyde, acrolein, and lipid peroxidation products. Risk assessment of exposure to genotoxic agents may therefore in the future assess the quantitative contribution of exogenous exposure to the endogenous background DNA damage, especially when the DNA adducts formed are identical to endogenous adducts. ^13,18,19 This certainly will need extended establishment of databases on background DNA damage in various segments of the population, including sensitive subpopulations. Once available, such data will provide promising novel perspectives for risk assessment, enabling assessment of the significance of DNA damage induced by a given exposure in relation to the physiological background.