Genetically modified organisms in aquaculture

Devin M. Bartley
FAO Fisheries Department, Rome, Italy

ABSTRACT

Aquaculture is recognized as one of the fastest growing food production sectors; part of the reason for this growth is genetic innovations and improvements. However, one technology that has sparked controversy in the areas of human and environmental safety is that of genetically modified organisms (GMOs). The controversy is most intense in the plant agricultural sector where 52 million ha in 13 countries around the world are planted with GM crops; 60% of all processed foods in the US are genetically modified. At present, there are no GM fish available to the consumer, however attitudes in the fishery sector are being influenced by events in terrestrial agriculture. Approximately 70% of the crops that are genetically engineered are engineered to herbicide resistance and many other crops are engineered to produce pesticides. Genetic engineering in aquaculture does not involve the engineering of toxins or resistance to toxins, but primarily focuses on improved growth rate; other traits of interest are improved environmental tolerance, sterility, and the production of pharmaceuticals. Although there are similarities between the agriculture and aquaculture sectors, significant differences exist in the area of genetic engineering that necessitate careful and focused review of GMOs in aquaculture. Science-based risk assessment and the precautionary approach have been widely promoted as tools for the responsible use of GMOs. However, consumers are influenced more by popular media than by scientific arguments; the aquaculture industry will be influenced mostly by consumer demands. Development of advanced genetic technologies such as GMOs will need to address both the rational and irrational concerns of the general public.

INTRODUCTION

Aquaculture is recognized as one of the fastest growing food producing sectors globally. Part of the reason for this rapid growth is the application of genetic technologies. Although there is a range of genetic technologies including conventional animal breeding, chromosome-set manipulation, hybridization, genomics, marker assisted selection, and genetic engineering, it is genetic engineering, also known as the production of genetically modified organisms (GMOs) or transgenic organisms that has generated the most controversy.

It is the purpose of this paper to present a brief status of the field of genetic modification in aquatic species and identify some of the important issues that will bear on their commercialization.

Status of Genetic Modification in Aquatic Species

Genetically modified organisms (GMOs) are defined by the European Union as "Organisms (and micro-organisms) in which the genetic material (DNA) has been altered in a way that does not occur naturally by mating or natural recombination". The technology is often called "modern biotechnology" or "gene technology", sometimes also "recombinant DNA technology" or "genetic engineering". It allows selected individual genes to be transferred from one organism into another, also between non-related species". The term transgenics is used when the technology specifically involves gene transfer from one species to another.

Although the methodology is complex, the basic idea is that a new piece of DNA called a genetic "construct" is made that is composed of a promoter, or switch, the desired gene to be introduced into the GM fish, a reporter or marker segment that allows the geneticist to detect the presence of the new gene, and a terminator segment that switches the gene off (Figure 1, Beardmore and Porter, 2003). Real gene functioning is much more complex that in Figure 1. For instance, the gene product may affect other parts of the animal and produce unintended results, the switches may also activate or deactivate genes other that the desired trans-gene, also producing unintended impacts, and the location of where the construct goes into the animals DNA is also unknown and may be inconsistent.

At present there are no genetically modified (GM) aquatic species available to the aquaculture industry. However, in the crop sector the technology is well established and is currently the subject of substantial debate. In the US, where the technology is generally accepted, approximately 60% of all processed foods are genetically modified. These mostly include products from soya beans, corn and canola. Around the world, approximately 130 million acres (52 million ha) in 13 countries are planted with GM crops. Gene technology is one of the fastest growing technologies, at about 15% per year. In the US, 3.5 million acres (1.4 million ha) were planted with GM crops in 1996 rising to over 88 million acres (35 million ha) in 2001. The majority of the genetic modification, about 70% of the crops, involves herbicide resistance, thus allowing farmers to control weeds without killing important crops. The remainder of the GM crops includes primarily plants genetically modified to express pesticides, thus allowing farmers to use fewer pesticides.

There are approximately 30 aquatic species that are being genetically modified in laboratories and test facilities. Some of these are being studied for commercial application, whereas others form the basis of basic research on gene and cell functioning.

TABLE 1
Experimental and developmental work on transgenic technology (GMOs) in aquatic species (after Bartley, 2000)

Table 1 lists some aquatic species being developed for genetic modification. There are two species of fish that are close to commercialization and awaiting government approval: 1) a transgenic Atlantic salmon in the United States and 2) a transgenic tilapia in Cuba. Both have been modified for improved growth.

ISSUES

The main issues involving GM fish are whether they present a danger to the environment or human health, and if they are ethical. Trade issues associated with GMOs are extremely complex and are not addressed here.

Environmental risk

The US National Research Council (USNRC) cited environmental issues as the greatest science-based concern for GM fish (USNRC, 2002). Their reasoning was that there are numerous uncertainties associated with how the genetic modification will affect the fish, how the GM fish will impact the environment and how the trans-gene may be passed to other populations in the environment. Further concern was raised because of the potential for GM aquatic species to escape from fish farming facilities and to spread easily and undetected through rivers and other water-bodies. The USNRC (2002) cited three main factors in environmental risk assessment: 1) the effect of the trans-gene on the GM animal - what change in the phenotype is expected and what are the potential unknown effects; 2) the specific animal, genetically modified - some animals have a tendency to become feral or invasive easily and have a history of causing environmental damage; and 3) the receiving environment - environments will be different as to their ability to withstand or recover from impacts of GM fish and environments will have different values to society.

Many of the environmental risks of GM fish are similar to those posed by non-GM fish. Any organism entering the environment can impact native biodiversity through predation/herbivory, competition, habitat modification and interbreeding with native species. Are these risks greater with GM fish? Due to a lack of adequate information from long-term and large-scale field studies, our ability to assess risks is not very good.

The ability of a GM animal to impact an environment will also depend on the specific environment. Ecosystems have different degrees of resistance to and resilience from adverse impacts. Resistant and resilient communities will be less impacted by GM fish than those unstable communities.

Many GM fish are modified to grow faster or have improved environmental tolerances (Table 1). If these fish escaped into nature, or if the gene or genes that confer these qualities on farmed fish were passed to native species through interbreeding, ecosystems could be adversely impacted through the increased activity (predation, competition, range extension, etc.) of the GM fish. Existing ecological balances could be offset by the introduction of a highly competitive or highly predacious GM fish. However, some researchers feel that because GM fish are domesticated and designed for life in fish farms, they will not be very competitive in nature if they escape (Dunham, 1999). There is still considerable disagreement among scientists on the issue of fitness of GM fish in the wild. Thus, case-by-case examination, experimentation and risk assessment are required at present.

In order to reduce the chance of escaped GM fish breeding in the wild, aquaculturists are examining techniques to make the fish sterile. This is easily accomplished in many aquatic species through adding an extra set of chromosomes (the creation of triploid organisms). Although this technique does not produce 100% sterile animals, and some males still exhibit secondary sexual characteristics, it greatly reduces the probability of fish breeding.

Health risk

Although most fishery regulatory agencies feel that environmental issues are of primary importance (USNRC, 2002), the human health concerns associated with GMOs in the human food chain^[5] receive a great deal of attention worldwide. This is in part due to news about crops. Crops have been genetically modified to contain pesticides, herbicides and general antibiotics, and there are fears that these toxins could affect people; the uninformed consumer feels that genetically modified fish also may contain toxins. An additional difference between plants and animals is that animals, in general, do not produce natural toxins or anti-nutritional compounds as many plants do. Thus, there is less scope in animals for the trans-gene to activate inadvertently naturally occurring toxins.

Although the risks to human health are slight, they are present and should be considered. GM fish could express genes or gene products, i.e. proteins, that do not have a history of safe use in the human diet. Risk managers are calling for a case-by-case evaluation of GM fish that first identifies the potential differences between the GM product and the non-GM product and then to identify the nutritional and toxicological implications of the differences. In evaluating food safety of GM fish, the DNA construct used to change the fish should be considered, especially if the gene or promoter comes from viral source. In this case, horizontal gene transfer or recombination (DNA combining with other DNA) could occur and lead to the generation of new viruses. DNA fragments may not be completely digested by the human gut and may survive in the gastro-intestinal tract. These fragments could be absorbed by gut micro-flora and somatic cells in the gut.

There have been instances in crops where the foreign gene has caused allergic reactions; for example, a gene from a Brazil nut was placed in soybean and people who were allergic to Brazil nuts reacted to the soybean. In the fisheries sector, the most common gene construct involves a growth hormone gene (Table 1) and not the herbicides or pesticides used in plants. Many of the GMOs being tested for use in aquaculture only produce more of their own growth hormone.

Labelling of GM products is currently a debated issue. However, many of the nutritional, toxicological and allergenicity concerns would be alleviated by such labelling.

Thus, from the human health perspective the risks are present, but minor. One area of potential concern is the future development of disease resistance. A theoretical possibility is that, if a GMO is more disease-resistant, it may become a host for new pathogens, some of which may be transmissible or pathogenic to humans.

Ethics and Animal Welfare

Is it ethical to modify genetically animals? In order to answer this question we must define what we mean by "ethics" and establish an "ethical framework" with which to evaluate the issue. A system of ethics is related to, but different from a system of values which is very much dependent on specific cultures. Many ethical frameworks exist or can be established, here I use a framework used in part by the FAO Sub-Committee on Ethics in Food and Agriculture. Components of this framework state that ethics included:

• Beneficence - towards people, animals and the environment. The use of GMOs that endanger people, harm or endanger animals, or have adverse environmental impacts could be judged unethical.

• Justice and equity - The use of GMOs or the development of technology that allows such use should be governed by ethical notions of justice - and fair treatment. This has been an issue with regards to patenting of genetic resources that have been used by traditional people, without their consent and without compensating them for the use of the resource. There is a feeling that the US Food and Drug Administration will soon approve a transgenic salmon for sale in the USA. However, there is also a feeling that the FDA will not allow the salmon to be grown in the USA! This could be judged an unethical decision because it does not treat the US and foreign environments in an equitable manner.

• Autonomy - the rights of individuals or groups for self-determination. This could also include the right to information to make informed choices. From the crop sector, control of GM seeds by multi-national businesses prevents farmers from saving seeds, forcing dependence on the multi-nationals and thereby reducing farmers' autonomy. Failure of labelling of gm foods also denies access to information that consumers need to exercise their autonomy in choosing GM or non-GM foods.

Ethical questions with regard to aquatic GMOs often focus on whether humans have the right to modify natural creations. Are we over-playing our autonomy? However, humans have been modifying plants, animals and the habitats they live in for millennia. The development of agriculture has been proposed as one of the most significant aspects of civilization in that it provided the time and resources that allowed humans to feed more people and left them free to develop fine arts and science. Genetic modification allows humankind to modify nature faster and to a greater extent than before.

LACK OF INFORMATION

From the above, it should be apparent that much of the controversy concerning the use of GMOs in fishery and aquaculture is due to a lack of information and scientific uncertainty. Major international agreements, such as the Convention on Biological Diversity (CBD, 1994) and the FAO Code of Conduct for Responsible Fisheries (FAO, 1995a), advocate a precautionary approach in such a situation.

The precautionary approach advocated by FAO and CBD states that where there are threats of serious or irreversible damage, lack of full scientific certainty shall not be used as a reason for postponing cost-effective measures to prevent environmental degradation. The Government of Sweden and FAO convened a technical workshop to define elements of the precautionary approach as they apply to fisheries (FAO, 1995b). These elements state that:

• reference points should be established to help determine desirable situations and undesirable impacts, e.g. limit reference points, such as lowest allowable population size of a fishery resource, and target reference points, such as optimum sustainable yield;

• pre-agreed actions or contingency plans should be implemented in a timely manner when limit reference points are approached, or when adverse impacts are apparent. Thus monitoring of the fishery, aquaculture activities and the environment is necessary to know when reference points are reached;

• priority should be given to maintaining the productive capacity of the resource where there is uncertainty as to the impact of development. This means that priority is given to conservation of stocks over harvesting the stocks when there is uncertainty. This can be extended to aquaculture where the productivity of aquatic resources in nature should be maintained when there is uncertainty as to the risk of GMOs adversely affecting them;

• the impacts of a development plan should be reversible within the time frame of 2-3 decades. The eradication of naturalized populations of GM fish, i.e. the reversibility of the impact, is difficult or impossible, especially in marine areas or in extensive river systems;

• the burden of proof should be placed according to the above requirements and standard of proof should commensurate with risks and benefits. The precautionary approach has often been taken to mean that the burden of proof rests with those proposing use or development of a resource, i.e. the aquaculture facility must prove that an introduced or newly domesticated species will have an acceptable impact. This is the "guilty until proven otherwise" approach.

CONCLUSION

GM technology could have much to offer the aquaculture industry, but it must be developed ethically with regard for the environment and human health. It is difficult to determine when GM fish will be commercialized and available to the consumer. Although scientists have established hazards to the environment and human health posed by aquatic GMOs, the likelihood and severity of these hazards is still unknown. GMOs offer the aquaculture industry additional opportunities to produce food and economic benefits. Improved growth-rate could mean less inputs and less waste from aquaculture; increased environmental tolerance could allow farming in marginal areas and provide additional employment opportunity; improved efficiency may allow fish farmers to move farms away from fragile coastal areas to areas less environmentally sensitive.

However, at present in some areas there is strong consumer resistance to the use of GMOs. GM plants and plant products are accepted in the USA and other areas, but not in Europe. Some question the need for GMOs and recommend traditional methods to achieve the same results. According to the United Nations, over the last decades food production has outpaced human population growth, yet approximately 800 million people are food insecure.

Part of the reason for consumer resistance is that there is a perception that GM technology benefits only the multi-national agriculture businesses and not the consumer. This seems to be the case with the plant sector as cost savings in GM food production were not passed on to the consumer. The crop biotech sector has strongly resisted labelling of GM products, causing further distrust of the technology by consumers. In general consumers do not know how their food is produced. In order for GM foods from aquaculture to be accepted, the industry must show how this technology benefits the consumer - is it cheaper, is it more nutritious, is it more environmentally friendly, is it more ethical? In the absence of clear and accurate messages on the benefits, consumers will base their opinion of GM fish on popular media and special interests groups. It will be necessary to produce science-based assessments of the risks and benefits of GM technology to consumers, policy makers and society in general.

REFERENCES

Bartley, D.M. 2000. Genetically modified organisms in fisheries. In: The State of the World Fisheries and Aquaculture. Rome, FAO. pp. 71-77.

Beardmore, J.A. & Porter, J.S. 2003. Genetically modified organisms and aquaculture. FAO Fisheries Circular. No. 989. Rome, FAO. 38p.

Dunham, R.A. 1999. Utilization of transgenic fish in developing countries: potential benefits and risks. J. World Aquaculture Soc. 30:1-11.

CBD. 1994. Convention on Biological Diversity. Text and Annexes. Interim Secretariat for the Convention on Biological Diversity, Chatelaine, Switzerland. 34p.

FAO. 1995a. Code of conduct for responsible fisheries. Food and Agriculture Organization of the United Nations, Rome, Italy. 41p.

FAO. 1995b. Precautionary Approach to Fisheries. Part 1: Guidelines on the precautionary approach to capture fisheries and species introductions. FAO Fisheries Technical Paper 350/1. Food and Agriculture Organization of the United Nations, Rome, Italy.

USNRC. 2002. Animal Biotechnology: science based concerns. US National Research Council. National Academy Press. Washington, DC. USA. (http://www.nap.edu/books/0309084393/html)

Government challenges facing food safety - industry perspective

David Rideout
Canadian Aquaculture Industry Alliance, Ottawa, Canada

ABSTRACT

There are a number of challenges that face government food safety regulators in the short and medium term. Reluctance by governments to deal with claims from those outside the system that some products are not safe, even though they meet government standards, is causing consumer confusion. Examples will be given in which the potential serious threat to consumer confidence in government systems will be highlighted. Possible solutions will be discussed from an industry perspective with specific attention to on-farm practices. As well, the risks to the food safety system caused by the current situation will be highlighted and it will emphasize the importance of the food safety system to industry and the potential diminishment of industry reliance on government standards if the status quo is allowed to continue.

AQUACULTURE AND THE CHALLENGES AHEAD

I am truly pleased to be here and to have this opportunity to speak to you about aquaculture from a Canadian perspective. Some of you are old friends and others are unfamiliar faces so let me start by telling you that age is a wonderful thing - it provides new insights into what before seemed like daunting tasks. I remember starting my work in fish inspection in 1974 and being questioned by friends who were looking to occupations much different from my chosen work in coastal communities. They highlighted for me that there did not seem to be any kind of a future in this obscure work called fish inspection that chance had put in my path.

I did not know what lay ahead as we do not know what the future brings. Looking back, we can see considerable changes. When I started my career in fish inspection, our challenge was to bring facilities and operations into the scientific realm - it was a hard fought task, but many processors today would shake their heads in wonderment as to why the issues in the 1970s were really issues. Sanitation and proper plant cleanup; construction of facilities so as to prevent bacterial growth; operational hygiene as fundamental as hand washing, these were all challenges. These challenges were taking place while visionaries were working internationally to develop standards and systems so as to ensure that country requirements were going to provide the best and safest products as well as ease of market access in fish trade.

Boats went out in the morning and came back in 12 hours or 12 days loaded with fish that was processed and prepared for market. Fresh, frozen, canned, salted and pickled, spiced and marinated fish was where we focused our attention. And, we carried a level of arrogance about how good our systems and standards were. Then along came HACCP. All of a sudden fish inspectors were no longer needed to verify every lot of fish. Daily sampling plans and holding room blitzes were no longer the way. Instead we were looking at new approaches centred on systems and system audits. Enforcement of the cops and robbers variety was replaced by industry due diligence and scientific auditing.

The fish inspection world changed forever and some of you were architects of that change. The result - consumers are provided with some of the safest fish ever at very competitive costs. However, some gaps in the HACCP approach exist and in Canada we are working overtime to eliminate those gaps. I am speaking here of on-farm food safety and traceability. I would argue that both are in practice in Canada but that they have not been codified. Given the consumer environment in which we sell our products it is essential that we complete this task of codification as soon as possible. I would argue that government should not be a passive observer in this process.

I was talking with an old colleague about some of the new programs and vulnerabilities of fish inspection systems. I was trying to figure out just how new and innovative processes were going to contribute to resolving issues. Not withstanding the tremendous progress in inspection systems in the last three decades, fish inspection regulators and the industry they regulate will be facing some of their greatest challenges in the coming months and years.

My friend and I were talking about aquaculture technologies and how public policy was not keeping up with the evolution of the Canadian aquaculture industry. Timeliness is essential to moving the industry forward. Unlike the wild fishery where the hunter-gatherer approach remains prominent, fish farmers need to plan much further ahead. Market projections for 2005-6 are now part of planning for the farming cycle which for some species can take 30-36 months or longer.

I recounted to my friend the old saying that life is a terminal disease, which is sexually transmitted. He then pointed out to me that I should not worry as we both had a 50% chance of living forever. While good odds, this was not a notion I had ever considered. Both he and I eat a lot of fish and we are all aware of the tremendous health benefits from fish consumption but... live forever? This was ground-breaking news. Forever, I said. Yes, he replied, today, half of the people who have lived on this earth are still alive.

All this to say that we are either going to face a population catastrophe or we will need to develop the means of managing our food production systems, and, I should add, our water as well, at a significantly enhanced level of sophistication. Aquaculture will definitely have a role in this evolution, but some would like to see it removed, or considerably scaled back from the food production process.

Physician and writer Arthur Conan Doyle once stated, and I quote, "It is a capital mistake to theorize before one has data. Insensibly one begins to twist facts to suit theories, instead of theories to suit facts". That is the situation facing fish inspection, and specifically fish farmers, today and it plays itself out in the mainstream media. I will try to explain, but I recognize at the outset that this is not an easy task without pointing fingers or contributing to a "he said, she said, we said, they said" debate.

We all know that research takes time. For the interim, we need to depend on existing knowledge and systems. However, in the quest to stop salmon farming, the anti-aquaculture lobby has gone after the hearts and minds of the public. In the absence of detailed science, facts are being twisted to suit theories. When we do receive more detailed science, we must ensure that any change it fosters is a result of reasoned debate respecting the entire food production system and the provision of the safest, most wholesome and nutritious products for consumers.

Anyone familiar with the value chain for food in general, and specifically aquaculture products, knows that the consumer is enjoying some of the best quality and safest fish products in history. However, over the past few decades there has been an increasing level of consumer concern raised by public interest and environmental nongovernmental organizations. Scientific detection levels are much more sophisticated now than they were twenty years ago and the media is devoting many column inches and time to health and food segments.

One need only look at the PCB issue related to salmon and you can see how easily a complex subject can be used to cause consumers to lose confidence and leave the fish counter all together. If the strategy of the environmentalists is to claim industry denial, it is the wrong strategy. We in industry recognize the results from studies that have been done and while we might question the methodologies, we feel sure that new more comprehensive and scientifically sound studies will show that there are low levels of PCBs in fish and that PCBs concentrate in fatty fish. However, the PCB levels will most likely be shown to be at least 40 times lower than the standards that North American food regulators set, as has been the case with studies to date. This is the nub of the debate - the suggestion that the standards are too high.

Let me make clear at the outset of this discussion that any PCBs in our products is a concern for our farmers. We in the salmon farming business in Canada fully understand the points raised by the environmentalists. We have operated under the correct assumption that our products were well within food safety guidelines, but we have also operated with a view to eliminating PCBs from our salmon. The new feeding technologies and efficiencies in feeding systems may make this possible.

The anti-salmon farming lobby argue that salmon represents an increased load of PCBs in the diet and that its consumption should be lowered through the application of a different standard. However, when we take the data provided by those claiming problems with salmon and extrapolate it from a "per serving" approach to a "per capita" approach, it becomes clear that salmon is not the source of PCB load when compared to foods like milk and beef. This is not an attempt to indict any other food or to dismiss the PCB levels in salmon but it does call into question the motivation of recent reports.

The question that challenges the fish inspection, and indeed the food inspection, community is whether these foods are in fact a food safety risk. I would say that they are not, but I do not think the public has been given a definitive view of what food regulators think. Is anyone in the food safety system telling consumers whether there is a food safety risk from salmon - no. Net result - consumer confusion, and public perception that leads to diminished confidence.

Worse is the fact that the consumer who moves away from fatty fish like salmon are removing one vital source of omega-3 fatty acids. Is there a risk here - you bet. From the 1997 US Statistical abstract we find that for every one million people in the United States, 2800 will die of heart disease. Eat fish, especially fatty fish and these risk levels fall. Silence by food regulators in instances where there is consumer confusion that causes those consumers to walk away from the fish counter could increase the risk to consumer's health.

I should point out that the evidence appears clear that if any of you have had a heart attack, you can reduce the risk of a second attack by 50% with a couple of meals a week of fatty fish like salmon as part of your normal diet. I could also discuss the benefits respecting brain, joint and skin from omega 3s found in fatty fish but I will leave that to another talk. My conclusion is that any silence by food regulators on issues like PCBs may have more dramatic effects than that of reducing sales of farmed and wild salmon. I do know that some of you have been speaking out, but the message needs to be more united and clearer if you want to get your message heard.

We can see similar examples in the area of pigmentation of salmon - pigmentation that mirrors in the feed what the animals receive in the wild. Read the environmentalist's mis-information and you will be told that farm raised salmon is injected with dyes. I know that that is just not true, but unfortunately as a paid employee of the salmon industry there are some who are sceptical of my motives for saying anything. But, the truth of the matter is that salmon are not injected with dyes. Further, the beta carotenes that are part of the feed of farmed salmon and that impart the colour, have also been part of the feed for hatchery salmon that make up a good part of the wild stock. The reason for its use in hatcheries is because of its positive health effects on the young salmon.

Silence by food regulators again leads to confusion on the part of consumers and a loss of trust in the food safety systems. Face it, in North America, we have one of the best food safety systems in the world - but ask consumers and you will find that in greater numbers, they are expressing a lack of confidence.

THE ROLE OF THE RETAILERS

One thing our research has shown us is that the consumer trusts their retailer and/or food service outlet. However, the "local grocer" is essentially non-existent in major North American cities and has been replaced by larger multi-outlet companies with rigorous consumer standards. If you were to ask me where there is rigor in the food safety system, I will increasingly point to the retail side, as they want to do everything possible to build and maintain consumer confidence and trust.

Retailers and foodservice operators are implementing measures that, in the past, one would expect from food regulators. And, talk about timeliness - when an issue arises, whether fact or fiction, if it is not refuted by the food safety system, we will see a certification requirement implemented by retailers within 24-48 hours. Let me point out here that I routinely get calls asking what is the government's public position on an issue, as the retailers are demanding answers from their suppliers.

We are now moving to a multi-certification system that goes against the efforts that you have been making in the area of equivalency and harmonization. We are doing this because of a reluctance to speak out about the level of risk to consumers. If you are looking to see the effects of the precautionary approach you only need to look at the policies of food retailers to see they will take no chances when it comes to their consumers. And, rightly so.

Failure to recognize this change will render much of fish inspection redundant and will be a source of significant frustration for fish packers who will need to meet a myriad of standards and codes established in the marketplace as a means of keeping consumer confidence and trust. I repeat that the question my members are asking is where is the government on this or that issue and the answer that I am providing is that they have "gone to ground".

"The death of democracy is not likely to be an assassination from ambush. It will be a slow extinction from apathy, indifference and under nourishment." I am quoting this statement made by Robert Maynard Hutchins, an educator, to underline the importance of maintaining strong systems and responding to accusations that bring the system into question. Failure to help manage the public perception of risk that is fed by the media can and will fundamentally change our system of governance.

Consumers need to make their choices based on the broadest base of information. The International Food Information Council goes one step further in stating that... "while consumer confidence in the food supply is relatively high, history demonstrates the vulnerability of food safety messages and messengers". This is not an easy environment that we are in. But, somehow we need to find ways of getting to the answers.

Consumers do not ignore the reports of a series of food safety incidents from around the world. They are looking for minimal risk and sophisticated systems are required to ensure that risk is managed and consumer illness and death from food is eliminated. Risk communication will be critical to providing overall system integrity. Having said this, risk managers are recognizing that there is a significant amount of consumer education required, as the best systems can be negated by cross contamination, poor refrigeration or unacceptable hygiene in the home. Media attention to food safety has created an atmosphere of cautiousness regarding safe sources of food and good choices for dietary nutrition. As consumer awareness develops, consumer confidence in relation to food safety is becoming increasingly precarious. This is heightened by the decreasing consumer confidence caused by sensational media headlines.

In Canada, my organization is working to develop state-of-the-art systems to ensure consumer confidence. We believe that one key element to resolution of the issues is for an industry driven on-farm food safety program, a national code system for sustainable aquaculture and effective traceability systems that can be validated through an industry based audit system. When problems occur, industry and governments must work together to foster the best science, relevant procedures and conception-to-consumption food management practices that, as much as is practicable, eliminate consumer risk and enhance consumer confidence.

Our efforts will not have the required effect without a co-ordinated and recognized approach to improving consumer confidence and responding to false claims or slanted stories that undermine consumer confidence. I think that we need to develop an action plan that gets us to recognized best management practices and standards for on-farm food safety and traceability covering all gaps in the entire fish safety system. We need to work together and with the environmental community to ensure standards will stand the test of time and contribute to a growing trust in the safety and healthy enjoyment of the variety of fish and fish products available to consumers today. Can this be done - from my perspective; the industry would welcome such an initiative and would work to make it succeed. My challenge is to the other food production industries and the scientific community to work together to reduce or eliminate contaminants like PCBs from our food through good manufacturing practices that are focused in these areas. But the real challenge is to food regulators to begin the process of dialogue and scientific exploration that eliminates issues that erode public confidence. These issues need leadership and the failure of governments to provide that leadership will be fundamental to the failure of the overall system. Action is required now.

Is this achievable - you bet. Will it happen? That rests with you folks, but I can say that for my industry we are working to make it happen. Failure will result in the evolution of a new system where food regulation will have a role but not one based on consumer confidence. We must all strive to make sure this does not happen.

Chloramphenicol, food safety and precautionary thinking in Europe

J.C. Hanekamp
CEO HAN, Zoetermeer^[6], Netherlands
G. Frapporti and K. Olieman
NIZO Food Research^[7], Netherlands

ABSTRACT

The authors evaluate the precautionary zero tolerance level approach to toxins in food from different perspectives, contending that as part of ensuring food safety, the precautionary principle is highly paradoxical and counterproductive. The chloramphenicol case is examined here, as a case in point. Availability biases, probability and system neglect in the application of the principle together with the regulatory 'moral free rider' dilemma are discussed. The authors conclude that the precautionary principle needs to be discarded from food safety regulations as it dramatically compounds the issue of risk and could moreover result in unlawful food safety regulations.

INTRODUCTION

The detection in 2001 of chloramphenicol (CAP), a broad-spectrum antibiotic, in shrimp imported into Europe from Asian countries was presented as yet another food scandal. The initial European response was to close European borders to fish products, mainly shrimp, from these countries and make laboratories work overtime to analyse numerous batches of imported goods for the presence of this antibiotic. Some European countries went so far as to have food products containing the antibiotic destroyed. This regulatory response spilt over to other major seafood-importing countries such as the United States.

The legislative background to their response is to be found in Council Regulation EEC No. 2377/90, which was implemented to establish maximum residue limits (MRLs) of veterinary medicinal products in foodstuffs of animal origin.^[8] This so-called "MRL Regulation" introduced Community procedures to evaluate the safety of residues of pharmacologically active substances according to human food safety requirements. A pharmacologically active substance may be used in food-producing animals only if it receives a favourable evaluation. If it is considered necessary for the protection of human health, maximum residue limits are established. They are the points of reference for setting withdrawal periods as well as for the control of residues in the Member States and at border inspection posts.

Additionally, Directive 96/23/EC ('the Residue Control Directive') contains specific requirements, in particular for the control of pharmacologically active substances that may be used as veterinary medicinal products in food-producing animals.^[9] This includes, primarily, sampling and investigation procedures, requirements as to the documentation for their use, indication for sanctions in case of non-compliance, requirements for targeted investigations and for the setting up and reporting of monitoring programmes.

ZERO TOLERANCE

It has been noted that existing legislation on pharmacologically active substances used in veterinary medicinal products significantly contributed to the decreased availability of medicines for uses in food producing animals in the European Community.^[10] One of the aspects not discussed in the EC reflection paper is the issue of zero tolerance, which has created a problematic situation in the international market. Council Regulation EEC No. 2377/90 contains an Annex IV listing pharmacologically active substances for which no maximum toxicological levels can be fixed. From a regulatory point of view any exposure to these compounds is deemed a hazard to human health. These substances are consequently not allowed in the animal food-production chain. So-called zero tolerance levels are in force for Annex IV. The reasons for this are obvious:

• The absence of an acceptable daily intake (ADI), and therefore an maximum residue limit (MRL), was understood as "dangerous at any dose" which "required" zero tolerance regulation;

• With the introduction of zero tolerance, a veterinary ban on Annex IV compounds (such as CAP) is required in order for, it was believed, the listed compounds to disappear from the food chain as veterinary use was given as the only source of the compounds;

• Earlier analytical equipment was not adequate to perform current tasks (limits of detection (LODs) developed from ppm (parts per million) to ppb (parts per billion) and ppt (parts per trillion)).

Thus, CAP - and other Annex IV substances - should not be detected in food products at all, regardless of concentrations. The presence of CAP in food products, which can be detected by any type of analytical apparatus, is a violation of European law and moreover deemed to be a threat to public health. As a consequence, food containing the smallest amount of these residues is considered unfit for human consumption. For all intents and purposes, zero tolerance is best understood as zero concentration. Only when Annex IV substances are completely absent from food (at zero concentration) the risks are deemed completely absent. The presence of CAP in food products is solely related to illicit veterinary use; other sources are not taken into account, or indeed considered, as they are not included in the legislation. Chloroform, chlorpromazine, colchicine, dapsone, dimetridazole, metronidazole, nitrofurans (including furazolidone) and ronidazole are the other compounds in Annex IV.

CAP is categorized by the IARC (International Agency for Research on Cancer) as probably carcinogenic in humans; group 2A.^[11] No acceptable daily intake (ADI) could be established for CAP due to the lack of scientific information to assess its carcinogenicity and effects on reproduction and because the compound showed some genotoxic (DNA damaging) activity.^[12] Within the regulatory context, this is understood to be as "dangerous at any dose".

The zero tolerance approach for Annex IV compounds applies the precautionary principle to food safety issues: "when in doubt, keep it out". The explicit goal of zero tolerance is not risk-based but precaution-based, as the absence of a MRL is from a regulatory point of view translated as "dangerous at any dose". Indeed, the European Community tries to uphold a high level of food standards to protect public health and safety. To that end the White Paper on Food Safety has been published.^[13] In this paper the Commission presents a number of principles to ensure a high level of human health and consumer protection, one of which is the precautionary principle. Although scientific knowledge is the buttress of European policy on food safety, the precautionary principle may be invoked where considered appropriate by the European regulators in view of the high level of protection deemed necessary. Zero tolerance is an example of invoking a precautionary measure.

Despite a ban on animal food production, CAP is still used in human medicine. It has a wide spectrum of activity against gram-positive and gram-negative bacteria. CAP therapy is usually restricted to serious infections when other drugs are not as effective. In The Netherlands a number of registered pharmaceutical products are on the market that are mainly used to treat eye infections.

OBJECTIVES

In this article, we want to evaluate the zero tolerance approach from different perspectives. As scientific knowledge on the exposure risks of CAP is limited, the precautionary principle, in the form of zero tolerance, is invoked. The regulatory attitude towards scientific data is typical of a precautionary culture in which a very high level of scepticism with regard to what science cannot do goes hand in hand with a very high level of confidence as to what science is supposed to deliver.^[14] It also shows the detrimental consequences of the European reluctance to bring together assessment and management in food safety issues; a theme we will succinctly address in the closing statements of this paper.

Our main argument here is that the precautionary zero tolerance approach, as part of ensuring food safety, is highly paradoxical and counterproductive. Availability biases, probability and system neglect in the application of the principle together with the regulatory "moral free rider" dilemma will be discussed here. Moreover, proof of no-presence and therefore proof of the absence of harm (zero risk) is implied in zero tolerance. This could well constitute a probatio diabolica, which will be discussed in relation to a recent Court of First Instance ruling.

The narrow focus on potential exposure risks, which are erroneously deemed to arise only from the illicit veterinary use of CAP, leaves a number of essential issues untouched. Firstly, CAP is a natural chemical, produced by the micro-organism Streptomyces venezuelae. Streptomycetes are a group of gram-positive filamentous bacteria belonging to the Actinomycetes, which are ubiquitous soil-bacteria found worldwide. The biomass per hectare of the Actinomycetes in 15 cm of topsoil is between 400 and 5,000 kilograms.^[15] Particularly members of the genus Streptomyces are well known antibiotic-producers.^[16] The first question that comes to mind is whether CAP could in trace amounts be present biologically in all kinds of different food products, thereby opening up a multi-source perspective not incorporated in present regulations? Might there be an ecological background for such antibiotics? As zero tolerance consequently translates into a best-available-techniques approach for analytical machinery (see below), this question is all the more pertinent, as increasing analytical capabilities could result in crossing this potential ecological boundary. We therefore embarked on a small survey, in which a number of European products, not typically related to the illicit use of CAP, were analysed for the presence of CAP. Below we will discuss the results and the related intricacies of analytical techniques and their progress.

Secondly, as already has been mentioned, CAP is still used in human medicine. Therefore, the environmental presence of CAP due to human clinical use needs to be looked at carefully. In particular, surface and waste water are targets of investigation, as they can become a source of CAP in food-production other than direct medication, adding to the multi-source issue. We will summarize the data that have been generated so far and will discuss their implications in relation to a zero tolerance approach to food safety.

Thirdly, the legal concept "zero" does not exist in the real world; zero tolerance effectively means a best-available-techniques approach in the quest for analytical limits of detection. Until the mid-1960s the general idea of food safety meant that food should not contain any potentially harmful residues of veterinary medicinal products. This was a more or less realistic goal because at that time residues could only be determined in concentrations of around 1 mg/kg (parts per million: ppm). Since then the availability and sensitivity of methods of analysis have continuously improved and the detection of concentrations as low as 1 ng/kg is common today. These improvements mean that ever lower amounts of residues are detected, which would previously have gone undetected. Efforts to enforce zero tolerance for CAP, but also nitrofurans and other antibiotics, have evoked international concerns for reliable analytical methods, regulatory harmony, practical modes of prevention and useful risk assessments. The sensitivity of analytical methods determines the operational definitions for "zero", and as the analytical sensitivities reach ppb (microgram/kg product) and ppt levels (nanogram/kg product), the cost of equipment and tests limit surveillance and furthermore increase the probability of detection. Below we will discuss intricacies of analytical techniques and their progress in relation to zero tolerance.

Fourthly, the risks of exposure to CAP through the food chain are regarded as dose-independent, meaning that any dose might give rise to disease. Indeed, protecting the general public specifically from toxic chemicals, particularly carcinogens, has been a principal goal of public policy. The REACH programme (Registration, Evaluation, and Authorization of Chemical Substances) is Europe's latest regulatory development in this field.^[17] In the absence of knowledge as to how a toxicant may harm individuals, regulatory toxicology assumes that even tiny doses can cause injury. This, however, is based on toxicological extrapolation models, which cannot be verified but serve as an axiom. There is in other words no proof whatsoever of the risks at low-level exposures; these risks are inferred through the linear non-threshold model (see below). Risk aversion has led legislation and regulation to seek to ban toxic chemicals or, if that is unattainable, to minimise exposure, for instance, to analytical limits of detection levels as is the case with zero tolerance. From a precautionary regulatory viewpoint the scientific impossibility of arriving at an acceptable daily intake, in the case of CAP for lack of data, is translated into "dangerous at any dose" or "no dose no cancer". The precautionary zero tolerance approach therefore is a regulatory interpretation of the linear axiom.

Two models to determine the dose-response relationship have traditionally been used in toxicology in the assessment and regulation of risks of toxicants: the threshold model (B) is used in the assessment of risks for non-carcinogens, and the linear non-threshold (LNT) model (A) is used to extrapolate risks to very low doses of carcinogens (Figure 1). The risks associated with low-level exposures to CAP are singularly inferred from the linear non-threshold axiom.

Calabrese and Baldwin, however, argue that the most fundamental shape of the dose-response is neither threshold nor linear, but U-shaped (C), and hence both current models provide less reliable estimates of low-dose risk.^[18] This U-shape is usually referred to as hormesis: a moderate stimulation of response at low doses and an inhibitory response at higher doses.^[19] It is to be regarded as an adaptive response of an organism towards toxicological perturbations. Acceptance of hormesis suggests that low doses of toxic/carcinogenic agents may reduce the incidence of adverse effects.

In Figure 1, tumours per animal are depicted on the vertical y axis, with the related dose on the horizontal x axis. The animal control group (not exposed to the carcinogen) is depicted by the black horizontal broken line at the 5-level on the y axis. The hormetic model C predicts a lower amount of tumours than the control group when exposure levels of the carcinogen are below 7 (on hormesis see the BELLE Web site [biological effects of low level exposures]).^[20] The hormesis concept challenges the axiom and use of low-dose linearity in estimating cancer risks, and emphasizes that there are thresholds for carcinogens. The particular choice of the LNT dose-response model in the assessment of the exposure risks of CAP and the role of the precautionary principle will be considered in the final section of this article.

THE RISKS OF CAP EXPOSURE

Aplastic anaemia (a form of anaemia when the bone marrow ceases to produce sufficient red and white blood cells) is the most dangerous effect produced by CAP. Its occurrence is extremely rare, albeit fatal and is only observed as a result of therapeutic treatment courses with CAP.^[21] The minimum dose of CAP associated with the development of aplastic anaemia is not known. Therefore it is unfeasible to determine a dose-response relationship for the occurrence of aplastic anaemia.^[22] Limited evidence exists for the carcinogenicity of CAP in humans exposed to therapeutic doses.^[23]

Nowadays CAP is only occasionally used for internal infections. Ophthalmic infections, however, are still treated with CAP. Documentation on the ophthalmic use of CAP provides no evidence that this route of administration is associated with the same toxicity risk as therapeutic CAP administered parenterally.^[24]

The available data on the genotoxicity of CAP show mainly negative results in bacterial systems and mixed results in mammalian systems. It was concluded that CAP must be considered genotoxic, but only at concentrations about 25 times higher than those occurring in patients treated with the highest therapeutic dose.^[25] Moreover, no adequate studies are available to evaluate the carcinogenicity of CAP in animals used for experimentation.

The total aplastic anaemia incidence estimated by the JECFA (Joint FAO/WHO Expert Committee on Food Additives) is in the order of 1.5 cases per million people per year.^[26] Only about 15 per cent of the total number of cases was associated with drug treatment and among these CAP was not a major contributor. These data gave an overall incidence of therapeutic CAP-associated aplastic anaemia in humans of less than one case per 10 million per year. In considering epidemiological data derived from the ophthalmic use of CAP, systemic exposure to this form of treatment was not associated with the induction of aplastic anaemia. All in all, there seems to be no evidence whatsoever that low-level exposure to CAP, either as a result of ophthalmic use or of residues in animal food, is related to aplastic anaemia.^[27]

When considering the difference between therapeutic exposure - as a result of which aplastic anaemia has been observed, albeit rarely - and exposure as a result of food residues - as a result of which aplastic anaemia has never been observed - it is clear that CAP does not present any hazard. The food residue exposure levels shown in Figure 2 are taken from the RIVM study (Rijksinstituut voor Volksgezondheid en Milieu; Dutch National Institute for Public Health and Environment) on CAP in shrimp.^[28]

The RIVM in their above-mentioned study estimated the cancer risk as a result of the consumption of shrimp containing CAP.^[29] The concentrations in imported shrimp varied roughly between 1 and 10 ppb (parts per billion; 1 and 10 mg/kg product). The estimated reasonable worst-case risk as a result of eating shrimp containing CAP is lower than the MTR-level by at least a factor of 5 000 (being a 1:1 000 000 added cancer risk in the human population).

THE ECOLOGY OF CAP AND ITS POTENTIAL PRESENCE IN FOODSTUFF

Of approximately 12 000 known antibiotics, it is estimated that some 160 are or have been used as human medication. The Streptomycetes account for well over half of these commercially and therapeutically significant antibiotics - the antibiotics are produced by means of complex 'secondary metabolic' pathways. Many other pharmaceuticals such as anti-tumour agents and immuno-suppressants are also derived from the Streptomycetes.^[30] A small sample of these Streptomyces derived antibiotics are presented in Table 1.^[31]

TABLE 1
Some Streptomyces antibiotics

The overall natural production of antibiotics by Streptomycetes under natural conditions is unknown. Nonetheless, it is possible to isolate CAP from Streptomyces venezuelae present in the soil.^[32] Considering the ubiquitous occurrence of antibiotic-producing Streptomycetes, it seemed interesting to investigate the potential natural presence of CAP in different kinds of food products not associated with the illicit use of the antibiotic. To that end the "Instituto Technológico Agroalimentario" (Agri-food Technology Institute: AINIA), an accredited Spanish, non-profit organization created by, among others, companies in the food-manufacturing sector, was asked to sample ready-to-sell products acquired from retailers for the presence of CAP. A commercial ELISA (enzyme-linked immunosorbent assay) kit for detecting the presence of CAP was used. A number of samples that presented a high value in the ELISA test were confirmed by HPLC-MS (High Performance Liquid Chromatography-Mass Spectrometry) technique.^[33] Of the total amount of food products tested (83 in total):

• 40 percent were below the ELISA LOD (limit of detection: 0.05 ppb [parts per billion]);

• 44 percent of the tested products gave a response between the 0.05 and 0.5 ppb;

• 16 percent responded above the 0.5 ppb;

• One of the ELISA positives was confirmed by HPLC-MS as containing CAP (the other HPLC-MS tested samples were below the LOD of 1 ppb).

The HPLC-MS confirmed sample concerned Spanish white wine with an estimated CAP concentration of 2.7 ppb. We will further discuss these results below.

PRESENCE OF CAP IN THE AQUATIC ENVIRONMENT

Pharmaceuticals (both human and veterinary), personal care products and other domestic organic contaminants have been detected in the aquatic environment (rivers and lakes). These contaminants are sometimes referred to as PPCPs (Pharmaceuticals and Personal Care Products). PPCPs comprise all drugs, diagnostic agents (such as X-ray contrast media), "nutraceuticals" (bioactive food-supplements), and other consumer chemicals, such as fragrances and sun-screen agents.^[34]

In other studies, emphasis is put on their point of entry in the aquatic environment -namely waste water and waste water treatment plant discharge - where the chemicals are referred to as organic wastewater contaminants (OWCs).^[35] However, the veterinary use of pharmaceuticals results in a diffuse dispersion in the (aquatic) environment comparable to, for instance, pesticides.

Focusing on antibiotic presence in sewage treatment plant effluent and surface waters, Hirsch et al. published the analysis of various water samples for 18 antibiotic substances.^[36] Interestingly, CAP was detected in the effluent of a sewage treatment plant in the south of Germany at a maximal concentration of 0.56 mg/l. In surface waters, CAP again was detected, at a maximum concentration of 0.06 mg/l.

OBSERVATIONS AND DISCUSSION

Zero tolerance as a precautionary regulation is intended to eliminate certain risks to human health as a result of exposure to residues in animal food products. This in effect means three things in relation to food safety: (i) detection of CAP as such, irrespective of concentrations, is deemed a public health risk, following the LNT maxim from which the risks are inferred (not observed); (ii) detection of CAP in food is singularly related to the illicit veterinary use in food production; (iii) a precautionary zero tolerance approach in food safety of the illicit veterinary use of CAP would "totally" remove CAP (and its concomitant risks) from the food chain and food products. The last two aspects will be discussed here; the first will be tackled in the conclusion.

CAP's usage as a medicinal antimicrobial and antibacterial agent could result in its release into the environment through various waste streams by which food may be contaminated during the production phase. Indeed, Hirsch et al. did find CAP in the aquatic environment.^[37] It was detected in the effluent of one sewage treatment plant and in surface-water at concentrations of 0.56 mg/l and 0.06 mg/l respectively. So, through human clinical use, CAP can enter the food chain. This fact alone makes the presumption untenable that by banning illicit veterinary use, a zero tolerance regime would eliminate its presence in food. In other words, with CAP we are dealing with a multi-source issue.

Hirsch et al. surmise that this environmental presence might also be due to the veterinary use of CAP, despite its legal status as being part of the Annex IV. However, from the large number of groundwater samples that were taken from agricultural areas in Germany, only two sites showed any contamination by antibiotics. More importantly, municipal waste water is usually not disposed of with animal manure from farms. This suggests that intake from veterinary applications to the aquatic environment is of negligible importance. As sales of CAP in Hong Kong are between about 11 times and 440 times greater than in several western countries and Australia, environmental contamination of surface waters as a result of human use is expected to be at much higher levels than in Germany and the United Kingdom.^[38]

The present knowledge on the spread and behaviour of PCPPs is still anecdotal and is biased towards "finding" the contaminant in the environment. However, the detection of PCPPs warrants an unbiased review of concentration levels of CAP (and other PCPPs) in the aquatic systems to assess and quantify its distribution. In designing PCPPs base surveys, experience of monitoring pesticides may be useful. Pesticides/herbicides/insecticides as a contaminant group are comparable because (i) they are also applied in small loads (although considerably higher than PCPPs); (ii) they are detected in very low concentration; (iii) in the aquatic environment they show a temporal concentration fluctuation over the year;^[39] and (IV) the chemical transport behaviour and the metabolites are often not very well known.

As CAP is a natural antibiotic, natural contamination of numerous food products is a definite possibility. This hypothesis was tested by means of a small survey described above. The results are difficult to interpret. They do not indicate proof of a measurable ecological background of CAP, despite its ubiquitous natural producer. The presence of CAP in white wine, however, does represent an interesting, albeit inexplicable, caveat for the possibility of a natural bacteriological source of CAP in the food chain, adding to the multi-source argument.

A problem with the AINIA data, as with all data produced at the edge of analytical limits of detection, is that food matrices artefacts to which the ELISA responds cannot be differentiated from a real presence of CAP. The reality of false-positives ("detection" of a non-present target-molecule) is a well-known problem in the analytical sciences.

Results obtained in a recent collaborative trial on the determination of CAP in shrimp provide an illuminating illustration.^[40] Together with a number of CAP-spiked shrimp samples in which predefined quantities of CAP were added to the shrimp, blank (unspiked) samples were also tested. In the blank shrimp samples, three laboratories of the 14 participating laboratories measured CAP levels of 0.27 (n=2), 0.42 (n=1), and 3.98 mg/kg (n=3). The last two results were marked in this collaborative trial as outliers (meaning that these results were not deemed to be valid). However, any given real-market sample would have been judged positive by these laboratories for the presence of CAP and removed from market. In this case each of the three laboratories used a different method: ELISA, GC-MS/NCl and HPLC-UV. Four other laboratories obtained results varying from 0.03 to 0.09 mg/kg, using ELISA (2x), GC-MS/NCl (1x) and LC-MS-MS (1x). These low values could again be designated as false positives. Alternatively, they could reflect a natural background level, as indicated by us in the AINIA results. In summary, 50 per cent of the laboratories designated blank shrimp as being positive for CAP, which are worrying results in view of present political unease.

The sensitive detection of analytes has improved dramatically during the past decades, including methods used for the detection of CAP. Screening methods based on immunochemistry showed on average a tenfold improvement of sensitivity for the detection of CAP in milk powder every seven years (Figure 2). The developments in instrumental methods, which are used for confirmation, have been less prominent. On average they needed 14 years to increase tenfold in sensitivity. However, it is not unlikely that the instrumental methods will show a more rapid evolution in the future. The sensitivity of LC-MS-MS has improved circa tenfold in the last six years and a fundamental limit has not yet been reached.

Only a very tiny amount of the analyte becomes ionised in the widely used electro-spray interface and again just a tiny amount of the ions produced in the spray is actually sampled by the mass spectrometer. Therefore, one can expect that detection of CAP in ng/kg (parts per trillion) will become feasible in the next decade. In combination with the present zero tolerance policy this will lead undoubtedly to the destruction of increasing amounts of food and feed due to contamination by the natural presence of CAP or remnants of it in municipal waste water after human medical use.

There is some confusion about the Minimum Required Performance Limit (MPRL).^[41] MRPL is the concentration level that regulatory laboratories in the European Community should at least be able to detect and confirm. The MRPL should not be mistaken for a tolerance limit, or any similar terminology. EU regulatory laboratories are therefore obliged to try and find residues of banned substances, like CAP, at the lowest technically possible concentration. As a result of that, depending on the skills and equipment of the laboratories, lower than MRPL concentration may lead to the result being positive ("non-compliant sample"). The policy of zero tolerance can and will lead to economic inequality: products designated as safe by an exporting country can be designated 'non-compliant' if the importing country uses a more sophisticated method of analysis, resulting in lower detection limits. In that sense the MRPL did not produce a harmonised market made more complicated by the zero tolerance issue.

CONCLUSIONS

The simple legal inference espoused by Council Regulation EEC No. 2377/90, that when a compound on the Annex IV list is detected in food products its presence is the result of illicit use in food production, is falsified in the case of CAP. The straightforward legal reasoning that detection of CAP in food can only imply illegitimate use does not hold and needs to be revised as CAP is a multi-source issue. The reality of the human clinical use of CAP resulting in a measurable environmental source of contamination of food products bears witness to that. The AINIA results might even be indicative of a natural source. The presence of CAP in Annex IV of Council Regulation EEC No. 2377/90 is therefore superfluous. Moreover, zero tolerance will become a legal artefact as a result of increasing analytical capabilities in which the possibility of false positives will continue to haunt the legal issues. Clearly, we do not surmise that CAP is never used illicitly in food production.

The false positives issue surfaced poignantly in the German trial discussed above. In 50 per cent of the labs, blank samples were found to contain CAP. As has been said, no distinction can be made between the possibility of false positives and the possibility of a background concentration level due either to environmental contamination through human clinical use or to a natural bacteriological source or even to illicit use. Correspondingly, the source of CAP, when indeed detected, will be more diverse than is covered by the Council Regulation EEC No. 2377/90.

In recent months, another multi-source example surfaced concerning nitrofurans, also listed in Annex IV. SEM (semicarbazide) has long been considered a characteristic metabolite of the antibiotic nitrofurazone. Studies have shown that the parent drugs (nitrofurans) are rapidly metabolized by animals and are therefore undetectable. The stable metabolites are, however, detectable for a number of weeks after application of nitrofurans and are therefore regarded as reliable indicators of the (illegal) application of nitrofurans. Evidence for illicit use is, therefore, related to the detection of the metabolites such as SEM.

However, recently, SEM was found as a contaminant in food packaged in glass jars, which was not related to the nitrofurans at all. SEM is formed by thermal degradation of azodicarbonamide (ADC). ADC is used as the blowing agent in plastic gaskets of packaging material. SEM migrates from the gaskets into food products.

SEM was also detected in special animal and vegetable matrices that had been concentrated using drying procedures, like heating to reduce water content. A substantial formation of SEM was observed after samples were treated with hypochlorite (bleach) in accordance with common food processing methods used for disinfection or bleaching.^[42]

The examples of CAP and SEM open up a broad perspective on numerous other multi-source cases whereby zero tolerance policies will, of necessity, fail as a means to remove certain products from the animal food production chain.

The choice of the LNT maxim to underpin zero tolerance is in line with the precautionary principle, which however holds a strong availability bias.^[43] The above-mentioned toxicological model is well known to the regulators and follows in a longstanding toxicological tradition, whereas other models are not, or are to a much lesser extent, known. Also, veterinary use as the one source of CAP is generally accepted to be fact by the regulators and is part of Annex IV of the specific Council Regulation whereas other sources are not recognised at all. Moreover, the linear model is an attractive one, as it proposes complete regulatory control over the CAP risks, whether or not these risks are relevant.

This last point brings us to another bias, namely probability neglect (meaning that the probability of an outcome is neglected).^[44] The precautionary zero tolerance approach is focused on the outcome of CAP exposure in humans - aplastic anaemia or cancer - and neglects the probability of this outcome. Worse, the envisioned outcomes - aplastic anaemia and cancer - are merely theoretically inferred on the basis of the LNT toxicological model discussed above, and not empirically observed.

The JECFA committee concluded that low-level exposure to CAP is not associated with the induction of aplastic anaemia. Realistically, the risks as a result of CAP exposure from food consumption are nil. Indeed, neglect of probability here leads to the probability of neglect.

The hormetic U-shaped model advocated by Calabrese and Baldwin openly and scientifically challenges the regulators' choice of the LNT model (from which, as said, the risks are theoretically inferred):^[45]

The a priori criteria we developed to assess whether experiments displayed evidence of hormesis based on study design, magnitude of the stimulatory response, statistical significance of the stimulatory response and reproducibility of findings, revealed up to 5 000 examples of hormetic responses independent of chemical class/physical agent, biological model and endpoint measured. Low levels of agents such as cadmium, dioxin, saccharin, various polycyclic aromatic hydrocarbons, X-rays and various gamma-ray sources reduce tumours in some species. Low doses of X-rays enhance life span in male and female mice and guinea pigs; ethanol and acetaldehyde enhance longevity in fruit flies; multiple stressor agents extend longevity in nematodes; numerous toxic substances (for example, cadmium and lead) enhance growth in various plant species. Low or modest consumption of ethanol reduces total mortality in humans, while increasing it at higher levels of consumption. The hormesis concept is thus highly generalizable and far-reaching.

Hormesis redefines our concept of 'pollution' and 'contamination'.^[46] It questions the premise that "pollutants" are unreservedly bad. This is revolutionary because modern environmental and public health legislation is built, in a large part, on the moral dichotomies of good versus evil, clean versus dirty, natural versus unnatural.^[47] Annex IV exudes "badness". Zero tolerance, and thereby zero risk, is the express goal of Annex IV and of many who advocate the precautionary principle.^[48] Hormesis challenges the very premises of the Annex IV list: things are not either bad or good; they are both, depending on exposure levels and adaptive responses from the exposed organisms. In our view the LNT maxim needs to be reconsidered in relation to its use in food safety regulation.

This brings us to our final reflections. When a single problem is examined, it can be difficult to see the full consequences of legal interventions.^[49] The precautionary principle has the appearance of being workable only because a limited subset of the relevant effects is "on screen". The key aspect of system neglect is the risk of trade-offs. This is especially salient in the light of international trade, as the zero tolerance approach has until now resulted in a trade-off (such as the faulty communication of risk)^[50] between perceived food risks for Europeans and economic risks of the exporting countries no longer accepted by Europe as trade partners. The risk-risk trade-off translates into a health-health trade-off to the detriment of the exporting countries.^[51] Moreover, with zero tolerance policies the reality of multiple-sources of the banned substances is ignored, the potentially ambiguous nature of the methods of analysis is not adequately tackled (such as in the case of SEM), and the issue of false-positives is not considered implying that the very basis of MRL legislation is in jeopardy. Additionally, the Second Law of Thermodynamics challenges zero tolerance policies, as zero concentration - as implied by zero tolerance - is not a physico-chemical reality.

Zero tolerance has, so far, resulted in the race for ever lower limits of detection. As a result, analytical technology becomes a goal in itself, irrespective of toxicological relevance of the concentrations detected. Food safety, as such, has been disregarded for a legal construct. Indeed, with zero tolerance, proof of absence of a banned product, and therefore proof of no harm is brought to centre stage. Zero tolerance stands, in other words, for zero risk. This, however, is a scientific impossibility. Indeed, in the Pfizer case on the antibiotic growth promotor virginiamycin, the Court of First Instance remarked:^[52]

130. Supported more specifically by Fedesa and Fefana, Pfizer submits that in any such risk assessment, the Community institutions must show that the risk, although it has not actually become a reality, is nevertheless probable. The existence of a 'very remote risk' should be allowed given the concrete positive elements arising from the use of the product concerned. In any event, the Community institutions cannot legitimately apply a test which Pfizer describes as a 'zero risk' test. Such a test is inappropriate since it is impossible to satisfy. It amounts essentially to requiring probatio diabolica from the industry, something which is recognised as unlawful in all the legal systems of the Member States (Opinion of Advocate General Mischo in the Greenpeace case cited at paragraph 115 above, ECR I-1651, at I-1653, point 72). It is never possible to prove conclusively that a chemical or pharmaceutical compound or anything created by modern technology represents a zero risk to public health now or that it will do so in the future. To apply such a test would quickly lead to the paralysis of technological development and innovation.

In the light of this ruling, Annex IV of Council Regulation could be considered as unlawful, as zero tolerance promulgates the explicit goal of zero risk, which is unfeasible in the real world and demands the impossible of economic parties. A way out of this predicament is that Annex IV should list only compounds that clearly show toxic effects at very low dosage. Proof of no harm is then rewritten in proof of harm; a much more solid base for regulation, which does not generate a probatio diabolica for industry.

The precautionary zero tolerance approach encourages people to think that "safe" food actually exists - which is an impossibility - and is, with the implementation of the precautionary principle, within reach. More importantly, with zero tolerance, chemical food safety is presented as the prime aspect of food safety as a whole, which is explicitly not the case. On a relative scale of risk, food safety issues rank as shown in Table 2.^[53]

By scrutinizing the chemical safety of food products, other aspects of food safety run the risk of receiving a lower priority in public and politics. Furthermore, such an approach to food safety carries the risk of intensifying the search for banned chemicals, as has been the case in Europe so far, tying up budgets, research efforts and personnel to the detriment of food safety as a whole.

TABLE 2
Ranking of food safety issues in relation to human health

The precautionary principle clearly belongs to the broader precautionary culture, which holds the view that society's "systems managers" have a duty to prevent all damage, irrespective of cost and reality.^[54] It encourages people therefore to become moral free riders by overlooking their own responsibilities. In that sense the European food laws contain a NIMBY (Not In My Back Yard) aspect through the precautionary principle, in which the view is propagated that potential public health risks, for instance as a result of low-level exposure to CAP, is to be averted at all cost. Cost-benefit analysis is often criticized for comparing the costs of some with the benefits of others. The precautionary principle, however, does not seem to be doing any better. In the case of CAP, affluent European citizens avoid immeasurably small potential risks with the result that citizens in exporting countries have to forgo very real economic opportunities with ensuing risks to the quality of their lives. Indeed, the perceived, albeit absent, benefits from zero tolerance for the European population are converted into economic and public-health costs of the exporting countries.^[55]

In this article, we have attempted to broaden the picture on the CAP issue and have covered many issues. A rational system of food safety regulation is certainly cautious in its review of risk, but at the same time needs to be aware of and take in the range of the issues at hand. Proof of no harm cannot and never will be a guide for food safety regulations, as this requires massive research efforts focused on minute risks. And, even then, the gathered data might again give rise to further questions, resulting in an endless scientific quest. A probatio diabolica indeed.

In our view, the case of zero tolerance and its failure to improve food safety demands a reappraisal of the strict separation between risk assessment and risk management. The assessment of risk, or the lack of it, has by definition policy implications that need to be addressed in order to avert mishaps such as described in this paper. The absence of an acceptable daily intake for CAP does not imply 'dangerous-at-any-dose' at all, as it only derives from a lack of data. A precautionary zero tolerance policy therefore is superfluous. Consequently, in our view, the precautionary principle needs to be discarded from food safety regulations as it dramatically confounds the issue of risk.^[56] It only addresses the perception of food safety as opposed to food safety itself.