Guide to Risk Assessment and Biosafety in Biotechnology, GRABB

An Initiative of the United Nations Environment Programme (UNEP)




 Tools of Regulation


 Julian Kinderlerer


Contact: IRRO Secretariat


Risk Assessment

Risk Perception

Risk management

Risks associated with Contained Use

Summary of Laboratory Containment Requirements

Risk assessment for the Release of Genetically Modified Organisms

Example 1: Crop Plants

Table 2: A summary of the information required to submit a proposal for the field release of transgenic plants in the European Community

Example 2: Transgenic Fish



 This chapter attempts to understand the methods which may be used for risk analysis and risk management for genetically modified organisms. Thetically modified organisms. The range of issues which need to be considered in order to consider the hazards which might be posed by the introduction of these new organisms is formidable. It remains difficult to identify all the problems which might have to be considered in order to perform an adequate risk assessment and therefore introduce a management programme, particularly where the organisms are to be introduced into the environment. Unlike chemicals, where it may be assumed that the effects will dissipate over a period of time, organisms may persist and increase in the environment, and the risk assessment must consider what the risks may be, and attempt to minimise damage.

However a chapter of this sort is cast, it puts the new technology in a spotlight which appears to be pessimistic -- it appears to be dangerous to allow the technology to happen. We are considering risks, not benefits, which may be very substantial. Even though the risks may be small, they have to be considered.


Although the appearance is pessimistic, the tenor is optimistic, for the result of risk assessment in most instances should be to define the required risk management procedures which will allow the work to proceed. In many (if not most) instances, the risk assessment would identify few risks. The fact that a risk analysis has been performed, and procedures instituted to ensure safety, means that fears of the new technology should be able to be allayed.

There is often a presumption that a risk assessment must be a complicated document which provides all sorts of nightmare scenarios. Where a risk assessment has been performed for the use of water as an aqueous solvent in a biotechnology laboratory, I have seen instances where the scientist has written down that a risk of drowning in the water exists. In a biotechnology laboratory, with millilitres rather than hectolitres of water, this is simply nonsense! This ham-fisted attempt discredits assessment of risk and results in contempt for a process a process that is extremely important to our understanding of the processes used in the laboratory. In many instances a risk assessment could be simply "in my judgement, based on experience, there is no risk associated with…." If an accident happened, then a court would have to decide whether the scientists judgement was indeed flawed, or was not really 'judgement' or whether at the time of attempting the risk assessment other persons with similar training would have made the same judgement

Risk Assessment

Where do we start when considering risk assessment for the use of genetically modified organisms? It is not clear what the risk might be, as has become obvious. If we are considering an organism which has become a pathogen to man, animals or plants then the risk is clear, but it is not simple to define harm to the environment. Where do we start?

We need a clear definition of risk first. Generally we attempt to separate two parameters, the hazard ase hazard associated with the modified organism and the likelihood of this hazard being expressed. Hazard is an intrinsic property of the organism; the polio virus is considered to be extremely hazardous, whereas brewer's yeast is considered benign and lacking in hazard under almost any circumstances. The likelihood of the hazard being expressed will depend on the way it is handled -- many highly dangerous viruses can be handled without risk in special facilities. Risk is then likelihood of the hazard being realised. Assessing risk is then simple in principle. The hazard has to be identified, or putting it another way, examining what in a particular situation could cause harm or damage and then assessing the likelihood that harm will actually be experienced, and what the consequences would be.

The risk assessment encompasses everything that has to be considered to decide about the hazards posed by the organisms and by the activity, or use of the organisms, the likelihood that they will actuallll actually give rise to harm, the control measures needed. The level of detail needed to perform a risk assessment will vary greatly depending on the circumstances. Most regulatory regimes require a "suitable and sufficient" risk assessment. For a simple operation involving a low hazard, well known and well understood organisms, and containment which is clearly suitable, the result of the assessment could be identified very quickly, and little needs to be noted except that the risk assessment was actually performed. For dangerous organisms about which there is uncertainty the risk assessment would obviously be extensive and may require new data before the work can proceed(1).

The Presidential and Congressional Commission on Risk Assessment and Risk Management in the United States has been examining a framework for Environmental Health Risk Management. In their reports, published in January 1997 and April 1997, they have produced a framework in which risk management will be conducted by the US GovernmUS Government in the future(2). They identify Risk to be the probability that a substance or situation will produce harm under specified conditions. "Risk is said to be a combination of two factors:

Risk encompasses impacts on public heath and on the environment, and arises from exposure and hazard. Risk does not exist if exposure to a harmful substance or situation does not or will not occur. Hazard is determined by whether a particular substance or situation has the potential to cause harmful effects"

Mackenzie and Henry(3) considered evaluating risk using the formulation

risk = exposure x hazardwhere exposure is a measure of the organism's (or the insert's) abhe insert's) ability to escape from the environment for which it has been designed. The subsequent fate of any escaped insert or organism then needs to be quantified in terms of the likelihood of persistence, increase and spread in the environment. Hazard is the impact of the organism of the existing ecosystem.

We therefore have a starting point, we first have to define the hazard associated with the modified organism, then look at the way we handle it in order to determine its risk. But our problem remains, where do we start in determining the intrinsic hazard of a newly manufactured organism?

All of the foregoing implies that it is possible to identify the hazard. The starting point for this is always the intrinsic ability of the organism to cause harm. In many instances we will have information -- we will know, for instance that the host organism is pathogenic, or there are forms of the organism which have been demonstrated as having adverse effects (producing alleducing allergens or toxins, for example). If we know the likely hazards which the organism may cause, we can consider each of them and identify the risk associated with the hazard, and the likely exposure of the target individuals or the environment to the hazard. Unfortunately, there is uncertainty regarding the estimation of hazard, let alone risk. There is often little data available on the hazardous properties of the biological agents.

In these instances we have tended towards a precautionary approach, where it is assumed that it is better to be safe than sorry regarding the exposure of people or the environment to the risk. "where there are significant risks of damage to the environment, the Government will be prepared to take precautionary action to limit the use of potentially dangerous materials or the spread of potentially dangerous pollutants, even where scientific knowledge is not conclusive, if the likely balance of costs and benefits justifies it"(4)

Risk Perception

Estimating the risk associated with the use of a modified organism is hard enough, as has been demonstrated chapter 1 and again later in this chapter, but the concern which the public or pressure groups within society feel about particular risks depends on many other factors as well. Social scientists have studied this, and Fischoff et al have shown that people rate risks according to "how well the process (giving rise to the hazard) is understood, how equitably the danger is distributed and how well individuals can control their exposure and whether risk is assumed voluntarily"(5,6),

It is clear that risks are perceived in an 'illogical' fashion. The risk of harm to an individual eating a genetically modified food is probably many powers of ten less than the risk of there being involved in a traffic accident while driving a car. Few consider the risks associated with driving before getting behind the steering wheel, yet research has indicatas indicated a fear of eating modified organisms.

Risk management

It has proved useful to make a distinction between an assessment of risks (the evaluation of the likelihood of harm and its consequences) and risk control (the prioritisation of risks and the introduction of measures that might be put in place to reduce, if not prevent, the harm from occurring)(6). This distinction results from a presumption that we are easily able to assess the hazard, and therefore the risk and can be scientific and objective. This separation is artificial, and risk management is often an integral part of the risk assessment, as an iterative approach needs to be taken to minimise the exposure to the hazard.

Framework for Risk Management

The framework is conducted

If the hazard is identifiable, and the likely exposure may be estimated, we can estimate the risk. Risk management is then possible. Risk management will identify procedures to use which minimise the risk by decreasing the exposure to the hazard or by decreasing the hazard -- changing the organism to one intrinsically safer. Risk management and risk assessment are closely inter-woven, as the risk management approach will, by definition, modify to risk assesodify to risk assessment to minimise risk, and an iterative approach to the management may become important.

Perhaps the most interesting of the points raised in the preceding boxed table is the penultimate point, the requirement to collaborate with stakeholders: It returns to the problem that the public perception of risk is not correlated with any logically derived risk analysis. The Presidential and Congressional commission is placing within the envelope for managing the risk a need to take into account the prejudices of scientists, developers and consumers.

Risks associated with Contained Use

It is easiest to consider the risk if we start assuming containment, for all procedures which finally result in a marketed modified organism will begin in the laboratory. In some senses this is where most of the hazard is likely to be found, or where risk potential is greatest, for the least is known about modified organisms in the research environment, and sment, and scientists are more likely to modify pathogens or try and identify genes which will be 'useful' from a variety of organisms which are not fully characterised.

If the organisms are contained we have first to consider only the probable effect on human health and assume that our containment system will isolate our organism from the environment.

The European Union defines "contained use" as being within physical barriers, whether or not supplemented by chemical or biological barriers to escape or release into the open environment. The definition includes the modification of organisms and their storage, culture, use, transportation or destruction for which physical barriers are used to limit their contact with both the general population outside the containment and the environment. A genetically modified large animal might be considered to be in containment, if the definition depended on an ability to recall the 'organism' in the event of a problem, rather than a physical fence between tce between that organism and the wider environment.

In order to assess the risk for contained use, it is usual to follow a scheme where we

* consider the predicted properties of the genetically modified micro-organism to determine if there are any potential mechanisms by which it could represent a hazard to human health.

*consider the likelihood that the genetically modified micro-organism could actually cause harm to human health

* assign the containment which would be necessary to safeguard human health

* identify any hazards to the environment and assign any additional containment measures to assure that the environment is not placed at risk.

Our modified organism is conceptually separable into the host organism, into which genetic information is inserted; the donor organism, from which the genetic information has been derived; the vector which shuttles the information between these organisms, and the insers, and the insert, which contains one or more genes which display biological activity. It is useful to consider each of these in attempting to assess the likely hazard posed by the modified organism.

When working in containment all cells, whether they are microorganisms, plant, animal or human cells, are considered to be micro-organisms when used in culture. This is only important in that the European Community Directive (EC 90/219) deals with contained use of modified micro-organisms. Animals and whole plants used in containment are not covered by a Europe-wide Directive, although many of the countries have "over-implemented" the directive so as to include all organisms used in containment.

In general, however, it is only microorganisms which are considered pathogenic to humans, although plant cells may produce toxic and allergenic substances which pose a hazard to the worker in the containment facility. (The concept of toxicity includes mutagenicity, carcinogenicity, neurotoxicity and enviry and environmental effects).

For each of the donor, host, vector and modified organisms we may consider the hazard they pose, which will provide information which allows a first approximation to the hazard likely to be posed by the modified organism.

Before considering the properties of each of these, we may consider whether it is likely that the modified organism differs from the host organism at all. It is almost certain that the genetic information transferred to the host cell is an infinitesimal fraction of that incorporated within the host cell. The gene or genes that are inserted are likely to be well characterised and the changes in phenotype are predictable -- otherwise there would be no point in the modification (except in the laboratory where making gene libraries provide the possibility of totally unknown mixtures of organisms). However, the genetic modification might affect the host range of the host organism, or its capacity to utilise a different set of metabolites, or might con might convert the host organism into a pathogen, or alter its ecological niche, or the balance of organisms within that niche. The point of insertion of the characterised genes within the genome of the modified organism is unknown. All that this paragraph says, however, is that we should treat our organism as a totally new organism, and base a risk assessment on encountering an unknown organism. Are the hazards unique to modified organisms, or should we treat this as equivalent to a newly isolated organism?

If we were to treat this as a newly isolated organism, we would be discarding a great deal of information which we possess! Where would risk-assessment start? The implication would be that the organism should be treated as a very dangerous pathogen until proved otherwise, and surely the insertion of a non-toxic gene into a species like Saccharomyces cerevisae which has never been shown to have pathogenic properties would be extremely unlikely to produce a very dangerous human pathogen.

We obviously start with information relating to the host species, take into account all the information available from the donor and vector to allow a preliminary risk assessment for the final modified organism. The risk management procedures that will then follow will include some monitoring to ensure that the risk assessment performed in this way is not seriously wrong.

The genetic modification might affect the host range of the host organism, its capacity to survive under different conditions, and the susceptibility to effective treatment or prophylaxis in the event of infection. It might also alter its capacity to utilise different substrates or alter its balance with other ecologically interrelated populations. Nevertheless, it is the pathogenic properties of the host organism that determine the starting point for our assessment of the likely hazard posed by the modified organism.

Infection followed by disease will depend on the microorganisms ability to multiply in the tiply in the host and on the host's ability to resist or control the infection. It has proved useful to categories all microorganisms into 4 groups which define their pathogenicity to humans; only the first group are non-pathogens. This categorisation applies only to the infectivity towards humans, and is of significance only, therefore, for the contained use of organisms:(7)

Hazard Group 1: Organisms that are most unlikely to cause human disease

Hazard Group 2: Organisms capable of causing human disease and which may be a hazard to laboratory workers, but are unlikely to spread to the community. Laboratory exposure rarely produces infection and effective prophylaxis or effective treatment is usually available

Hazard Group 3: Organisms that may cause severe human disease and present a serious hazard to laboratory workers. They may present a risk of spread to the community, but there is usually effective prophylaxis or treatment available

Hazard Group 4: Organisms that cause sevthat cause severe human disease and are a serious hazard to laboratory workers. They may present a high risk of spread to the community, and there is usually no effective prophylaxis or treatment.

The intention of this categorisation, which applies to non-modified organisms as well, is to identify appropriate containment which would be required to protect those working with the organisms. The higher the hazard group, the greater the containment required to control the organism and ensure that it does not infect those working with it.

Pathogenicity is not a simple characteristic. Many genes must interact appropriately for a microbe to cause disease. the pathogen must possess and express characteristics such as recognition factors, adhesion ability, toxigenicity and resistance to host defence systems. Single gene modifications of organisms with no pathogenic potential or history, or even the introduction of multiple genes unlikely to confer pathogenicity are unlikely to result in unanticipated picipated pathogenicity. For example, E. coli K12 has been disabled to remove some of the factors that might be associated with pathogenicity (wild type E. coli is a group 2 pathogen). The factors which have been lost include the cell-surface K antigen, part of the LPS side chain, the adherence factor (fimbriae) that enable adherence to epithelial cells of human gut, resistance to lysis by complement and some resistance to phagocytosis.(8) This variant of E. coli is a common host organism for genetic modifications within the laboratory.

The starting point for the risk assessment is, therefore, an assumption that the level of risk associated with the modified organism is at least as great as that of the host organism (until proved otherwise, either by direct observation, or by argument where the factors which are likely to enhance or decrease pathogenicity are considered as in the case of K12 above). Whether in the laboratory or in industry the capacithe capacity to choose a host means that in all but a few cases the host organism will have been chosen to be in hazard category 1. It is assumed that the modified organism will be used under the same containment as the host wild-type organism unless the modification inserts information which would alter the pathogenicity.

The vector has also to be considered, both for its own potential for pathogenicity and for its ability to transfer the insert to other than the intended organism -- horizontal transfer of the information. Most vectors used for E. coli contain no sequences which might result in pathogenic behaviour. The presence of genes coding for antibiotic resistance might be of concern, but for most of these the antibiotic resistance is already so common in the environment that it can be discounted.

Most common E. coli vectors are transfer deficient, but the ability to transfer information either directly or with the assistance of other plasmids and the host rathe host range of the vector must be taken into account when considering the safety of the mechanism of insertion of the required genes into the host organism.(9)

The properties of the insert are again of importance in considering the risk assessment for the modified organism. Clearly if the information encodes a toxic gene product, or one which is known to be likely to modify the pathogenicity of the organism into which it is inserted, the great the risk. If the gene product is non-toxic and is not one which may pose a risk to the people working with the organism in containment, the risk management will largely be based on the pathogenicity of the host organism.

In most instances the characteristics of the donor organisms are of less relevance to the risk assessment than those of the host. If the donor organism is merely used as a source of well characterised DNA for a selectable phenotype or a promoter or other control sequence, the characteristics of the donor are unimportae unimportant to the risk assessment. If however, the insert contains genes which are biologically active, toxins or virulence factors, then information from the donor organism are of consequence. The construction of cDNA or genomic libraries make it essential to consider all the possible hazards associated with the donor organism, and in this instance, the hazard group may well have to be the higher or the two within which the host and donor fall.

It is now possible to examine the modified organism and consider the likely risk. During the 1970's Dr. Sidney Brenner and others in the United Kingdom attempted to systematise the approach by considering three factors -- Access, Damage, and Expression. The approach was incorporated in the United Kingdom's approach to risk assessment for contained use of bacteria, and is discussed in detail in a document produced by the Advisory Committee on Genetic Modification in the United Kingdom. The latest version of the guidance was publishas published in 1993 and provides clear guidance as to the risk assessment for the contained use of genetically modified microorganisms (including any cells in culture). The guidance note is free and may be obtained from the Health & Safety Executive in Britain. More information is available by looking at the newsletters published by the ACGM which are available on the internet on

Access is a measure of the probability that a modified micro-organism, or the DNA contained within it, will be able to enter the human body and survive there. It is a function of both host and vector. Depending on the organism being used, there are a number of routes of entry which allow access. The properties of the vector, particularly mobilisation functions need to be taken into account. In general if the organism is capable of colonising humans then access is high, whereas if the host is disabled so as to require the addition of specific nutrients not available in humans or outside of tside of the culture media and is also sensitive to physical conditions or chemical agents present in humans, then the access factor is likely to be low.

Damage and expression are usually associated with the insert and the gene product.

Expression is a measure of the anticipated or known level of expression of the inserted DNA; if the 'gene' inserted is intended to be expressed at a high level, for example, by deliberate in-frame insertion down-stream of a strong promoter, expression is likely to be high. If the insert is simply there to allow probes to detect the DNA, and is non-expressible DNA, i.e. with no foreseeable biological effect or gene containing introns which the host is incapable of processing, then the expression factor will be low. Examination of the final product, the modified organism itself, will determine the actual expression, which may be higher or lower than expected.

Damage is a measure of the likelihood of harm being caused to a person to a person by exposure to the GENETICALLY MODIFIED MICRO-ORGANISM, and is independent of either expression or access. It is associated with the known or suspected biological activity of the DNA or of the gene product. The activity of the organism which results in any toxic, allergenic or pathogenic effect need be taken into account within this parameter. It may be that the biological activity of a protein is dependent on the host cell system in which it is expressed. An oncogene expressed in a bacterium will have no discernible effect, when present in a human cell, problems may arise. The full biological function of many gene products require post-translational modification which will not occur within a bacterial cell normally. The potential biological activity of the gene product should be considered in the context of where an how it has been expressed and the effect on its structure and activity of the mode of manufacture. The range of 'damage' might be from

Once an estimate of each of these parameters has been made (in the United Kingdom this is numerical in steps of 10-3), they may be combined. The result provides a qualitative measure of the risk, and allows a containment level to be assigned for the use of the organism in order to protect those working with the GENETICALLY MODIFIED MICRO-ORGANISM.

Unfortunately, this Brenner scheme is only easily applicable to a small class of experimental uses of modified micro-organisms, but the number of experiments in research laboratory environments which fit the requirements for the application of this scheme make its retention useful.

Modified organisms may be used in containment in laboratories (or pilot plants) or may be used in an industrial setting. It may be that the primary distinction here is not the size of plant or type of organism, but rather the skill and training of those working in the facility.

It is likely that a research or develsearch or development laboratory will be working with organisms which pose a greater threat to either the individuals working therein or to the environment than do those organisms developed for large scale factory use. The great majority of organisms used in industrial production are well-characterised, 'familiar' organisms capable of being used under conditions of 'Good Industrial Large Scale Practice' or GILSP. Given that it is usually possible to 'choose' the parental organism into which a gene is inserted for a particular 'industrial' purpose, there would be no good reason to choose an organism likely to pose problems to either those working in the facility, or to the environment in the event of an escape.

The same logic would apply to the development stage where 'industrial' use of the modified organisms is being planned. There is a possible extra hazard in that it is at this stage that the modified genes may be inserted into the organism, and the unpredictability of insertion site may, arguably, arguably, require slightly greater care than that taken at the production facility.

In the research laboratory, organisms may be pathogenic to humans and/or to the environment, as it is here that fundamental research would be conducted. Experiments will involve organisms and /or inserts which may be injurious to the health of the workers or to those who are incidentally on site in the laboratory.

There are usually four levels of laboratory containment (based on WHO classifications) which are defined for pathogenic organisms (not just for genetically modified organisms) and are based on the hazard classification described elsewhere.

Level 1 is the lowest level or containment, and requires no extra precautions above those required for good microbiological practice. In general, this means that(12)

The conditions for higher levels of containment, where the organism is considered a pathogen, are listed in the guidance note (see footnote ). For level 2, the major addition are the need to ensure that access to the laboratory is restricted to those needing to enter; that there be adequate space for each worker (at least 24 m3); an autoclave must be readily accessible and all waste materials must be made safe before disposal either by autoclaving or by incig or by incineration.

Table 1

Summary of Laboratory Containment Requirements(13)


Level 1 Level 2 Level 3 Level 4
Laboratory suite: isolation no no partial yes
Laboratory: sealable for fumigation no no yes yes
Ventilationinward airflow/ negative pressure Optional Optional yes yes
through safety cabinet no Optional Optional no
mechanical: direct no no Optional Yes<>no Optional Yes
mechanical: independent ducting no no Optional Yes
Airlock no no Optional Yes
with shower no no no Yes
Wash handbasin yes yes yes yes
Effluent treatment no no no yes
Autoclaveon site no no no no
in suite no yes yes no
in lab: free-standing no no Optional no
in lab: double-ended no no no yes
Microbiological safety cabinet / enclosure no no Optional yes yes
Class of cabinet / enclosure - class I class I/III class III

There will be risks to the environment for contained use of modified organisms as well. The escape of modified organisms used in laboratories should be of little significance as they should have been disabled so that, in the event of escape, they will be unable to survive. However, the assumption has been that the only concern is risk to human health and safety. The risk assessment has been predicated on this, and the possibility that the organism may be a danger to other organisms within the environment has not been fully considered in the discussion. The concept of the 'environment' which includes land, air and water as well as other organisms and humans, is so broad when compared to the enclosed environment of the laboratory that is it difficult to define a clear stedefine a clear step-wise approach to risk assessment. The risk assessment must try to consider all possibilities of what could go wrong, and attempt to ensure that these cannot happen, largely through the design of the organism being used. Methods for retrieving the situation should an organism escape from containment become important, and need to be planned at the outset rather than relying on the containment procedures to work. This is particularly important where the organism is a Level 1 organism as it will not infect humans, but if it escapes could be disastrous to plants, insects or other animals.

The extra hazards that need be taken into account so as to assess the risk should an organism escape from the laboratory environment include (a great deal of that which follows was included in Chapter 1, but is felt to be important enough to be repeated here):

The potential to survive, establish and disseminate within an environment distant from that in the laboratory. This may include the dispude the displacement of other organisms, or the modification of the ecosystem so that other organisms disappear or are replaced. Indirect effects may follow the establishment of a new organism.

Pathogenicity to animals and plants is clearly significant. The characteristics of the host organism and the modified organism which are relevant to pathogenicity, toxicity, virulence, allergenicity, colonisation, predation, parasitism, symbiosis and competition need be considered. If the host is pathogenic, the modified organism may be to a greater or lesser extent, and this should be considered.

The potential for transfer of the genetic material should also be considered. Conjugative plasmids, transmissible vectors or transposable elements which could contribute to the undesirable spread of genetic material between the GENETICALLY MODIFIED MICRO-ORGANISM and other organisms must be considered in the risk assessment.

The products of the gene expression which might be toxic to organisms other thsms other than humans needs consideration, or where the precautions taken in the laboratory to protect humans from the toxic effects are no longer present in the environment. An organism that has the potential to cause negative effects on other organisms as a result of an inserted gene coding for a toxic product will pose a hazard.

The organism may have the potential to cause negative effects on other organisms and these should be considered.

The loss of a gene in the organism is not a hazard in itself, but such instability may lead to the incorporation of the genes in other organisms which may result in harm to the environment. Again this must be considered, even though escape is not expected.

Large scale use of modified organisms in containment is different from use in the laboratory in a number of ways. In the first instance, it is almost certainly true that the organisms used in development or for industrial and commercial use are non-pathogenic. -pathogenic. They are generally used under conditions of 'Good Industrial Large Scale Practice' (GILSP) defined by a working group of OECD.(14)

The hazards posed by large-scale fermentation of genetically modified micro-organisms are of the same nature as for other biological agents, in particular(15)

There is nothing intrinsically more hazardous about the large scale use of genetically modified organisms in containment other than the potential for a greater degree of exposure to an organism and its biologically active products or the possibility that workers in an industrial plant are less skilled at handling biological material than laboerial than laboratory workers. In general, large scale users may well have chosen the best characterised host organism, as knowing the conditions under which the organism is likely to thrive makes industrial use more cost-effective.

The criteria for organisms to be used under Good Industrial Large Scale Practice conditions include

Organisms which are used to manufacture biologically active chemicals will obviously not fall within the definition of GILSP in many circumstances. There are containment approaches which need to be taken to handle these organisms which are identified in ACGM Note 6.

Risk assessment for the Release of Genetically Modified Organisms

The risk assessment where release into the environment is projected is totally different from that for contained use. The wide nature of the environment is such as to make it extremely difficult to produce a 'flow-scheme' for an assessment of risk. What almost all countries have done is to produceis to produce a series of questions regarding the organism, vectors and predecessors and the site into which it is to be released. These questions allow a case-by-case estimate of risk, based largely on familiarity with the organism, the host, donor and vector.

Because of the difference in concept between organisms designed for use in the wider environment and those designed for use in containment, the legislative requirements are often very different. In the European Union a different Directive (90/220) applies, and the approach is completely different to that described above. The Contained Use Directive assumes that all use will be within a single member state, therefore, the National Competent Authorities are permitted to extend the scope of the provisions within the Directive. In the case of Release or Marketing of a product (even where the product is to be used in containment), the "single market" within the European Union becomes important, and the provisions of the Directive must be immust be implemented insofar as possible in an identical manner in all member states.

In many cases, particularly with the release of genetically modified plants, the major difference is that the modified organism has been designed to survive in the environment, and in particular circumstances has been designed to be fitter for the environment than the wild-type. Examples might be heat-or cold tolerance, drought or saline-tolerance in plants, and tolerance to heavy metals or organic chemicals for both plants and microbes. If a fish is given a growth hormone gene to enable faster growth during early life, it may remove all food for other fish which develop more slowly and be more fitted to its new environment than the traditional fish. This implies many different approaches to risk assessment than that used for contained use.

The risks to the environment resulting from releases of genetically modified organisms will be analogous to the risk from release of classical biological control agents, or agents, or to the release of novel organisms to a new environment. Many of the same ecological, cultural, political and economic pressures will be present. There is much information about the effects of releases (of both modified and natural). Several 'lessons' have been identified(16)

The assessment of risk may well be different for crop plants released into the agricultural environment, and for microbial products which are released into the more general environment, or even where they are released into a polluted environment where the organism has been modified to use one of the pollutants as its sole carbon source. Many believe that risk assessments do not provide sufficient assurance to allow the release of genetically modified fish

The approach to risk assessment has been precautionary. There is not enough information to allow much generalisation of the risk assessment, and the app, and the approach used is to gather as much information as possible and then, on a case-by-case basis, attempt to formulate a risk assessment. The information requirements are very similar in all OECD countries, and are listed in Table 2. The issues are discussed in some detail in three useful references, books on Biosafety edited by Tzotzos and Persley & Giddings(17) and the paper by Tiedje et al(18).Basically, the data required will include:


Two example systems have been chosen, and discussed in considerable detail in order to identify the major elements of risk assessment:

Example 1: Crop Plants

Crop plants are grown in an artificial environment, the so-called agricultural environment, but escape into the 'natural' environment may be a concern.

Historically crop plants have not been subjected to formalised risk/safety analysis or risk management as there is a mass of knowledge, understanding and experience of the procedures for managing the introductions of crop plants developed by a wide range of breeding methods(20). In traditional breeding, crops are improved by cross pollination between plants with desirable characters, followed by selection of progeny with new gene combinations. Improvement by plant breeding methods is possible when the genes controlling the characters of interest are found within the crop species itselop species itself or within species that are sexually compatible with it. Various techniques have been used over the history of traditional plant breeding to increase the choice of genes available through traditional breeding and now include embryo culture, ovary culture and protoplast fusion. Even when a novel hybrid plant can be obtained there may be failure of chromosome pairing, or the genetic recombination necessary to introduce the foreign genes into the crop species.

One of the principal hurdles that have prevented the routine modification of crop plants, is the difficulty of introducing foreign DNA into a plant cell. The process of DNA introduction into plants (called transformation) was first achieved in tobacco, and has been relatively easy to achieve in other solanaceous species (potato, petunia, various Nicotiana species). In certain other crop species, especially the cereals, it has been less easy. Many approaches to transformation have been attempted attempted and the successful methods used at present fall into 4 groups. It is a digression to briefly discuss the principle methods for transformation of plants, but they do provide some information useful in risk-assessment:

The Agrobacterium method of transformation is used widely to transform dicotyledonous species. There are two principal species. Agrobacterium tumefaciens and A. rhizogenes which in their wild type form are pathogens causing crown gall disease and hairy root disease, respectively. Many dicotyledonous species are susceptible to infection by Agrobacterium species brought about by the incorporation of genes from an independently replicating plasmid within the Agrobacterium cell which then become incorporated into the host plant. The introduced DNA modifies the phytohormonal levels within infect cells and either causes a disorganised proliferation of cells and the formation of a gall or the production of a mass of roots covered with root-hth root-hairs.

The disease causing genes are carried between specialised T-DNA (transforming DNA) border sequences on the independently replicating circular plasmid DNA molecule. By recombinant DNA methods, it has been possible to remove the disease causing genes, so the Agrobacterium organism is no longer pathogenic. New vector plasmids have also been constructed which enable foreign genes to be inserted between the T-DNA borders. Agrobacterium cells carrying the gene(s) of interest are then incubated with cultured cells of the recipient crop plant and transgenic plants regenerated from them. Only a small proportion of the treated plant cells eventually become transformed, so it is usually necessary to incorporate selectable marker genes (usually conferring resistance to a particular antibiotic) between the T-DNA border sequences. To select the transgenic plants, the corresponding antibiotic is incorporated into a plant regeneration medium on which only transgenic plants are able tare able to grow normally.

The major restriction to the use of this technique is that many plants, particularly cereals, are extremely difficult to transform because of the lack of a wound response.

Protoplasts are plant cells that have had their cell walls removed by enzymatic treatment. They can be produced from various parts of the plant (often from leaves or hypocotyls) and are bounded by the plasma membrane. This membrane is delicate and its integrity can be affected by polyethylene glycol treatment or by passing an electrical current through a protoplast suspension. The DNA to be introduced is added to the medium surrounding the suspended protoplasts and the chemical or electrical treatment allows DNA to enter. In a small proportion of protoplasts the foreign DNA becomes incorporated into the cell genome.

As with the Agrobacterium method, it is usual to use an antibiotic resistance selectable marker gene in order to select the transformed protoplasts and the cell colothe cell colonies that develop from them. Plant tissue culture procedures are subsequently used to regenerate whole transgenic plants.

The Agrobacterium and protoplast methods have often proved inadequate for the transformation of recalcitrant species. Cereals are not normally hosts to Agrobacterium, and routine and reliable regeneration of plant from cereal protoplasts is difficult and often genotype dependant. The technique of particle bombardment was developed in an attempt to overcome some of these problems and involves coating metal particles (usually tungsten or gold particles, 1 m in diameter) with DNA and shooting them into plant cells capable of subsequent plant regeneration. Particles can be propelled by various means, including 0.22" (inch) blank cartridges, compressed gasses or by the instantaneous evaporation of a water droplet caused by electrical discharge. The small metal particles, with their DNA covering, enter the plant cells and become lodged there. In a tiny pro tiny proportion of the recipient cells the DNA becomes incorporated into the genome, and transgenic plants can be regenerated from them. As with the other methods, it is usual to incorporate a selectable marker gene.

Partial digestion of cells in multicellular structures: A method which now looks promising for the transformation of cereals, is the partial digestion of immature embryos with enzymes, followed by stimulation of DNA uptake by exposure to an electrical current (electroporation). Eventually plants are regenerated from the transformed cells of the embryo. This method has the potential advantage of using immature embryos which often have a high capacity for plant regeneration. More experience with this approach will be required before it can be established how widely applicable it may be for the transformation of a range of gramineous species.

What are the hazards posed by modified plants? -- Unwanted attributes of crop plants may include the tendency oftendency of a self-pollinated line to outcross because of self-sterility or other factors. There may also be a tendency to become a weed. The modified plants may produce toxic substances in the product or the target range of the toxin deliberately inserted into the modified plant may differ from that of the donor organism. The response of the modified organism to other organisms in the environment and the reaction of other organisms to the modified plant may change. Any of these may pose a risk to either humans working with or consuming the products, or to the environment. Where pesticides are introduced into the modified crops, the actual response may be much wider than that expected, with consequences for the ecosystem. In addition, the modified plant may display unwanted changes in appearance, susceptibility to environmental stress or end-use characteristics. "In many cases, the effects have been scale dependent and, therefore, became apparent during the scale-up process"

It has proved virtually ivirtually impossible to formulate a 'quantitative', structured method for the assessment of risk to the environment resulting from the deliberate release of a modified plant. For example, it is not thought possible to estimate the probability of a plant becoming a weed when released, as the characteristics of 'weediness' are not easily defined . "(21). The complexities of the natural environment and ecological interactions mean that risk estimation is, in most cases, more a question of qualitative evaluation rather than of quantitative analysis". Risk assessment procedures for the release of transgenic plants therefore require a detailed comparison of the transgenic plant with the plant genotype it was derived from. The procedures also require consideration of the interaction of the modified organism with the particular environment into which it is to be introduced. Although the methodologies in various countries differ, most ask similar detailed questions about the modified organisms and the environment in aent in attempting to perform a risk assessment. The system used by most countries requires a response to a very large number of questions relating to the organism, release site and wider environment. The information required is essentially the same in all countries. It involves a number of steps which identify the hazards associated with an introduction into the environment . (22)

An example of the information required for risk assessment is given in Table 2. The method of presenting the data varies from one authority to another, but the information required is essentially the same. In 1995 the Organisation for Economic Co-operation and Development assessed the processes within member countries for the "commercialisation of agricultural products derived for modern biotechnology".(23) 23 countries responded to their survey:

The categories outlined in Table 2 will need some explanation. The process involves the identification of possible hazards associated with the modified organism, the environment into which it is to be released, and the interaction of organism and environment. Hazard having been identified, it is possible to assess both the risk of that hazard occurring and the magnitude of harm if the hazard is realised. If the hazard is small and the probability of it happening is high, the risk to the environment remains small.


Table 2: A summary of the information required to submit a p to submit a proposal for the field release of transgenic plants in the European Community (24)

(This provides the basic information for risk assessment)(25)

General information

Name and address of the organisation wishing to release transgenic plants, including the names and qualifications of the personnel responsible.

Information about the DNA donor organism, the recipient plant species and the transgenic plant.

Characteristics of the transgene donor organism(s) and of the recipient plant species, including:

Characteristics of the gene vector used to introduce the transgene(s) into the recipient plant species, principally:

Characteristics of the transgenic plant including:

Information about the conditions of the release and the receiving environment.

A description of the proposed release, including:

 A description of the release site and the wider environment, including:

Information about the interaction between the transgenic plants and the environment

Characteristics of the transgenic plant may affect its survival, multiplication and dissemination

A description of the interaction of the transgenic plant with its environment, including:

An assessment of the potential environmental impact, including:

Information on monitoring, control and emergency response plans

A description of monitoring techniques, including:

A description of methods for controlling the site, including:

A description of methods of discarding waste plant material

A description of emergency plans to removed or destroy the transgenic plant material and to terminate the eterial and to terminate the experiment if it is considered necessary

 General information:

It is considered important that the staff and the institution responsible for carrying out the release have a sufficiently high level of expertise and experience to carry out the proposed release of transgenic plants and to be responsible for any field containment and monitoring that is considered necessary.

The DNA donor organism, the recipient plant species and the transgenic plant:

It is essential to have information on the donor organism and the recipient plant species. Information on the recipient species will establish a baseline against which to compare the transgenic plants. Knowledge of the donor species will highlight the kind of information required from the transgenic plant. If the donor is a plant pathogen, for example, this will raise questions in the riskons in the risk assessment exercise about the possibility of recombination between the integrated DNA from the pathogen and pathogens that may infect the transgenic plant subsequently. The possibility of transcapsidation may also need to be considered where the sequences inserted code for a viral coat protein to confer resistance to certain viral diseases. When the modified cell is infected by another virus, transcapsidation of the virus with the coat protein of original donor virus might be possible and this may, in turn, affect viral host range.

Transformation Vector:

Information is required on the DNA vector used during the transformation process employed to introduce the transgenes. Antibiotic resistance genes are generally used to facilitate the screening of transformed cells. Other DNA sequences may act as linking sequences with the vector or may provide other functions associated with the use of recombinant DNA methods. With certain transformation sysrmation systems carrier DNA is sometimes used to aid the transformation process. It is therefore necessary to know the nature of this DNA so that any consequences can be considered during the assessment process.

Transgenic Plant:

It is important to give a description of the transgenic plant including molecular data on the inserted transgenes, the stability of expression, whether there is any change in allergenicity, toxicity and the capacity of the transgenic plant to persist in agricultural habitats or invade natural habitats. It is essential here that the corresponding unmodified plant genotype is used as a control so that changes in plant phenotype caused by the transgenes can be measured.

The conditions of the release and the receiving environment: Although the scientific or commercial purpose of the release may not have any consequences for risk, it is important that the biosafety groups charged with the assessment oassessment of risk have perspectives against which to assess the release. The risk to the environment requires qualitative judgements and, therefore, an essential part of the risk assessment philosophy is case by case analysis and that based on the accumulated experience there is a progression towards streamlined and simplified procedures where appropriate (see conclusions). Providing information on the objectives of the release, its size and design and the agronomic treatments to be used are important both for risk assessment of the particular release and for the longer term national and international learning process.

Ecological information on the release site environment is also important. This should include a survey of plant species that might be growing in the vicinity of the release and information on what is known about the nature of pollen dissemination and the distances over which pollen can give successful pollination.

The location and type of the anticipated target organisms mustanisms must be specified. The target organisms are those which the transgenes are targeted to affect. For transgenes containing the insecticidal Bt protein it may be a particular class of insect pest, for a viral coat gene it would be a particular viral pathogen. Non-target organisms are those which are not the primary target of the modification and include those that are affected inadvertently. If the transgene has an effect on an insect that is not considered to be a pest, this should be noted, There should also be a consideration of whether the transgenic plant becomes a better or worse host and/or harmful to organisms that might be associated with the crop. The risk of harm to the environment includes harming non-target organisms.

The interaction between the transgenic plants and their environment: In order to determine the impact of the transgenic plant on its environment it is important to describe changes in the transgenic plant that may change its invasiveness in wness in wild habitats, its persistence in agricultural habitats or changes in its ability to propagate itself sexually or asexually. It is also important to take a note of earlier studies with similar transgenic plants. It is also necessary to determine the possibility of the transfer of transgene to the same or related plant species (wild and cultivated) or to micro-organisms and if this is possible what the consequences of that gene transfer might be.

Monitoring, control, waste treatment and the emergency response plans:

Once plants are released from containment and particularly if allowed to flower and set seeds there is the possibility of plants, seeds, pollen carrying the transgenes being transferred out of the immediate release environment. An important part of risk assessment is to determine the extent to which it is possible to monitor transgenes after the release and the efficiency with which it is possible to destroy pladestroy plant material if it becomes necessary. Efficient methods of identifying transgenic plants or transgenes in species they may have transferred to may be necessary. This may be by a visual marker (e.g. beta-glucuronidase), a selectable marker (e.g. antibiotic resistance) or by molecular analysis e.g. PCR and Southern hybridisation.

There are ways of minimising genetic exchange which might be considered (see later). It may also be appropriate to describe ways in which plant material can be destroyed at the end of the release experiment or if considered to be necessary during the course of the experiment.

Carrying out the risk assessment: The first releases of modified plants have been experimental. Commercial releases are now happening as the results of the experimental release have been analysed. It is at this experimental stage that the major risk assessment will have to have been done, and evaluated. The aim of the risk anthe risk analysis is to identify either changes to the experimental protocol or methods by which the GMO may be confined in order to minimise risk to the environment or human health.

Assessing risk is not an exact science. It is difficult to put a value on the degree of risk. It is never possible to establish that releasing a transgenic plant will have no risk. All of the activities people are involved in pose a degree of risk, no matter what kind of precautions are taken. The essential feature of risk assessment is to determine how the transgenes might alter risk compared with the non-transgenic crop, hence the starting point must be the use of the unmodified crop plant as a baseline against which to compare the effect of the inserted transgenes. There will undoubtedly be questions that cannot be answered because the relevant data are not available. For instance it is not known for certain whether plant genes can be transferred to micro-organisms, and there is much we do not know about the nature an nature and consequences of gene flow from our conventionally bred crops to related weed species. Biosafety committees have to take into account both knowledge and ignorance to arrive at a decision that on the acceptability of a release and where this is not the case, what additional information or precautions may be necessary.

In cases where detailed scientific knowledge is not available, it is important to use the experience of conventional plant breeding to aid the risk assessment process. Plant breeding has been carried out (first the result of serendipity and later of intention) for thousands of years and many of the genes being inserted by recombinant transformation fall into classes very similar to those manipulated by the conventional plant breeder.

One of the major concerns associated with the release of modified organisms is that the inserted information may be transferred to wild populations. The process of introgression (26) is of concern to many authors as a mechanism which may which may lead to undesirable traits being transferred from modified organisms. Inter-specific hybridisation is a ubiquitous process, notwithstanding the barriers that exist to cross-breeding, but most hybrids are rare and the majority are sterile. 'Gene Flow'(27) is believed to be highly restricted , but there is some evidence that this may be misleading. Modified plants could, in theory, become weeds difficult to control, possibly in contexts other than their normal agricultural environment.

A transgene may escape from a crop if the transgene is transferred to another crop or to another, wild related crop (possibly by introgression) and the plant containing it persists after the crop on the agricultural land, in verges, ditches or waste-tips or if the plant invades semi-natural habitats. It is likely that the spread of transgenes can be used using similar methodology to any other single gene trait.

What then, are the risks associated with the release of modified organisms?(28) Is it possibs it possible to confine the modified organism within the released site, and if it were to 'escape' would it pose a problem for the environment? Could it, for example, survive or persist outside the managed 'site' within which it had been released? There are differences in the potential of crop plants to transfer from the environment in which they are placed, and in their ability to establish feral populations. If this happened, would it matter? Could the inserted genes be transferred to other plants of the same type or to wild relatives?

The risk assessment must take the host or parental plant as its starting point. Is the host plant capable of surviving outside the normal agricultural environment? Does it have relatives in the external or agricultural environment with which gene transfer is possible? The modification must then be considered, both in terms of making any resulting transgenic plant more likely to survive and the safety of the gene product in the environment.

Some areas that nareas that need to be considered are outlined below as examples.

Antibiotic resistance genes are used as they meet all of the above criteria. The nptII gene which confers resistance to the antibiotic kanamycin is most commonly used. It is an antibiotic which has been superseded for clinical use in the USA. The resistance marker also gives resilso gives resistance to the antibiotic neomycin, which is still prescribed in some countries for clinical and veterinary use. There are also other aminoglycoside antibiotics that are also used as markers. Although several extensive studies have concluded that the likelihood of the transfer of this gene from transgenic plants to micro-organisms is negligible and probably of little consequence if it did happen (29), there is a continuing debate about whether it is acceptable for this kind of selectable marker gene should be present in commercial transgenic varieties.

Alternatives to antibiotic resistance markers have been used. These include herbicide resistance markers. There have been a number of suggestions for replacing antibiotic marker genes in plant systems. The Cre-lox system involves the production of two sister lines of modified plants, one containing a gene coding for a bacterial recombination enzyme, the other containing the antibiotic resistance gene flanked by sites for the action of the n of the enzyme. When the two lines are crossed, the marker gene should be excise. It is believed that this system would be difficult to apply in vegetatively propagated food crops, such as potatoes.

Gene-fusion markers are used to confirm the expression of proteins within the inserted gene, rather than for selection of the modified organism. One of the most commonly used markers is the ß-glucuronidase gene from Escherichia coli

A range of herbicide tolerance genes have been introduced into various crop plants. Herbicide resistance was one of the first traits subject to genetic modification as the mechanisms of resistance had been characterised. In general the resistance is a dominant single gene trait. One of the principal attractions for this application is that it provides a means of providing selectivity for herbicides that are quickly degraded in the environment. An environmental risk that needs to be considered is whether the transgenic crop plant that is herbicide tolericide tolerant may become a weed that is then difficult to control. Another factor that needs to be considered is the likelihood of the herbicide resistance genes becoming established in weed populations by hybridisation between crop and weeds. If the hybrids are fertile, they may be difficult to control in an agricultural system which depends for weed control on the same herbicide, or in adjacent crops that depend on the use of the herbicide for weed control. If there are other adjacent populations of the same crop which have been modified to be tolerant to different herbicides, would a crop plant resistant to multiple herbicides pose extra problems within either the agricultural or natural environment?

Were the herbicide tolerance to be transferred to non-managed, non-agricultural species within the 'wild' environment, would there be cause for concern? Such environments are not normally subject to herbicide treatment and the presence of wild relatives of crop species displaying resistance may be of may be of little significance. The risk assessment should attempt to identify the possible consequences of such a transfer.

Transgenes conferring pest or disease resistance could potentially confer a selective advantage on a crop plant and make it more persistent on agricultural land and more invasive in wild habitats. The transgenes could similarly confer a selective advantage if transferred to related wild plant species. Another aspect that needs to be considered is the effect of the resistance genes on pest and pathogen populations. If the transgene provides a very efficient defence it is possible that the pest or pathogen will rapidly become resistant. This is a phenomenon that is well known in conventional plant breeding and is arguably more about devising a sound agricultural strategy than assessing risk, but the possibility of using the same resistance gene in a range of different crops by transformation means that this prospect has to be taken seriously.

One of the most important usmportant uses of this technology for the insertion of genes leading to pest-resistance has been the use of Bt toxins. A problem that may be associated with the use of Bt toxins is the evolution of resistance in the target pests. This resistance is due to reduced affinity for the toxin to a mutant membrane receptor. Transgenic crops containing proteinase inhibitors may pose similar problems, and their use should be carefully planned to avoid the evolution of resistant pests.

Viral coat protein genes are frequently used to give protection against particular viruses. There is currently a debate about whether this strategy might modify the host range of plant viruses by a process of transcapsidation. Transcapsidation events have been detected in mixed infections of luteoviruses in the field. This process does not in itself create a new virus, as the coat-protein genes are unaltered, but may temporarily alter the specificity. There are also concerns over the recombination between the viral sequences expreces expressed by the plant and those of the infecting virus. The concerns will relate to the presence of other viruses capable of transcapsidation within the receiving environment.

Resistance to stress conditions such as drought, saline soils, heavy metals, cold, high temperature are complex characters so the isolation of genes conferring enhanced stress resistance will take time. Plant genes induced by stress have been identified (e.g. heat, cold, salinity, heavy metals) so it is likely that plants with enhanced tolerance will become available in the next few years. Transgenic plants of this type will present a particular challenge for risk assessment because this change may enable plants to grow in habitats where they were unable to grow before and may confer a selective advantage on plants to which the transgenes may transfer by cross pollination.

The exotic species model where a plant is transferred to a different country and become a dominant species is sometimes used to illustrate whatstrate what might result from the release of transgenic plants. While this is not a good general model for assessing the consequences of inserting one or a few genes into a crop to modify it in very specific ways, it may be relevant to instances where plants are modified to make them grow in new kinds of habitats. The results of a risk assessment in these cases may be that more information is required on the nature of any competitive advantage conferred by the transgene under the specific conditions of the release.

The gene product may be toxic in the plant or parts of the plant or it may modify the allergenic properties of the crop for food or products in the environment (e.g. pollen). Toxicity may apply to organisms other than the target organism, and harm may be to the ecosystem.

The risk assessment required when considering the widespread use of a transgenic crop variety are essentially the same as for small scale releases, but there are some important differences. Field containment measinment measures including those to prevent flower production and to destroy all plants material on the release site (field plot) are no longer possible, the risk assessment must therefore take account of the possibility of cross pollination between transgenic crop and adjacent non-transgenic crops and weed species. Transgenic plants will have the opportunity to become established in a range of natural habitats and there will be the opportunity for recurrent migration of pollen containing the transgenes into wild populations. In habitats where the crop plants vastly outnumber the sexually compatible wild species, the transgenes may become established and those wild populations even if they confer a selective disadvantage compared with the wild species counterparts not carrying the transgene.

There will be the opportunity for transgenic plants to be taken intentionally or inadvertently to other countries, including to those geographical areas that have a different spectrum of sexually compatible weed sble weed species. A small scale risk assessment considers the distribution of sexually compatible species in the location of the release site large scale releases raise the question as to what geographical limits should be placed on the environment considered in the assessment scheme.

Rare events which might not be evident in small scale release may be significant on a large scale. For example transgene instability on a small scale be unacceptable in crop varieties used on a large scale.

There are also questions of agricultural strategy, for example, how far is it prudent to progress towards the introduction of several different herbicide resistance genes into a crop species? Will this give the possibility of multiply resistant weeds? Also, what will be the consequences of the same or similar pest or disease resistance genes being present in the many different crops. The experience of conventional plant breeding may make it more likely for resistant pests and pathogens to emerge.


Other scale dependent questions regard the toxicity and allergenicity of the crop and the nature of the breakdown products of the transgenes when the plant decays.

The opportunities for managing risk in small scale releases are principally concerned with achieving an acceptable level of field containment and of monitoring the site during the release and afterwards. Increasing containment can be achieved by the inclusion of buffer crops, cages or bagging flowers in the field. Permission for the first release experiments in the United Kingdom required the de-flowering of all plants in the experimental plot. While ensuring that there the probability of transferring the introduced gene to other plants was low, this procedure could only be applied to small scale field trials.

The options for containment following widespread release of a transgenic crop variety are limited and a risk assessment before the commercial release of a transgenic variety must take thiust take this into consideration. It is important that the risk assessment leading to commercial release is thorough and open to scrutiny and considers all the evidence available from the small scale releases carried out with the same and similar transgenic plant material. As mentioned in the previous section, there may be difficulties associated with large scale releases. Some mitigating action might be considered for large scale releases.

There is now a wide range of plant promoters available for giving transgene expression in specific tissues in the plant. It may be desirable to restrict their expression of a particular protein to the parts of the plant where it is required. A pest that attacks leaves, for instance, may need an insecticide protein expressing only in leaves. There may also be opportunities in the future to use a constitutive promoter to give expression in most parts of the plant, but to use a strategy to switch off transgene action in very specific tissues, for example in pollen wh pollen where there may be the possibility of an allergic response.

The use of male sterile transgenic plants may be considered desirable to prevent the dispersal of transgenes in pollen.

There may also be instances where it is possible or desirable to grow transgenic crops in areas where no sexually compatible wild relatives grow naturally.

In order to prevent genetic contamination, it may be occasionally desirable to grow a particular type of transgenic crop (e.g. oilseed crops with a particular fatty acid composition) in areas free from non-modified crops.

When considering crop plants, it is not only their impact on the environment which needs to be considered, but also that they may be intended to be used for animal feed or human food. In addition, when the crop is out in the field, the farm-workers may come into contact with the plants or pollen at high concentrations. The possible toxic, allergenic, teratogenic or carcinogenic properties of the modified plants has to lants has to be considered as part of the risk assessment. In order to determine if the gene product new to the modified plant is allergenic sequence epitope homology with known allergens, heat stability, sensitivity to pH, digestibility by gastrointestinal proteases, detectable amounts in plasma and molecular weight may be considered. The allergenic potential of the donor and of the recipient organisms should be considered.(30)

Example 2: Transgenic Fish

The modification of animals is very different from that of plants. We are unable to regenerate animals directly from a single cell, so that the bombardment of a tissue as used in transforming plants is not available. The generation time of an animal (particularly large animals rather than insects) is too long to allow selection techniques which allow the transformation of a very few cells to provide a significant source of transgenic animals. For large animals the transformation must result in a very high success rate, probaate, probably more than 50%.

Although modification of animals is very different from that of plants, the questions which have to be considered in order for a risk assessment to be performed are very similar.

Techniques for transforming animals include microinjection, where the DNA is injected directly into a cell or the use of retroviruses which should result in stable integration of the genes within the genome of the transformed cells.

One group of animals which has proved controversial is fish, both because they may prove relatively easy to modify, and because they are not easily contained. Recommendations on marine releases made by the International Council for exploration of the Sea (ICES) are published in the "1994 Code of Practice on the Introductions and Transfers of Marine Organisms". The discussion which follows is largely based on a publication by the British Department of the Environment entitled "Guidance for the experimental release of Geelease of Genetically modified fish" referenced in footnote .

If fish are modified for aquaculture the likelihood of escape into natural ecosystems is thought to be high, and the fish might well carry traits which might result in significant impacts on ecosystems. In 1993 there had been experiments on 22 species using approximately 40 different DNA constructs.(31) Most of the work has been on

A major problem when compared to plants is the generation time for species of fish used in aquaculture -- generation times may range from less than one year for Tilapia to a few years for salmyears for salmonids.

"Microinjection into the fertilised egg is the most commonly used technique for introduction of a DNA construct. Unlike mammalian eggs, it is generally not possible to visualise the pronuclei or nucleus of fish eggs, and microinjection is generally into the egg cytoplasm before the first cell division. This often leads to high levels of mosaicism, with integration of one or more copies of the introduced DNA often occurring after the first cell division". Other techniques which have been used include electroporation of eggs and sperm, or soaking of sperm in the DNA construct prior to fertilisation. Biolistic techniques have also been attempted.

The hazards which could be identified for which likelihood would have to be considered include(31,32)

knowledge of the molecular biology and physiology of fish is limited compared to crops or mammals

The basic science 'database' for fish is limited; although the potential for modificl for modification for use in aquaculture is vast, detailed knowledge particularly of the physiology of most fish species is limited


If the GM fish proposed for release has very specific habitat requirements the distribution of the fish after release may be able to be predicted, and monitored. If the fish roams over a wide variety of habitats, the capability to monitor will be greatly decreased. The impact and survival of the modified fish depends on their ecological fitness relative to wild stocks. The performance of these fish, however, could be very different from that observed in containment. The relative fitness could be unchanged, increased or decreased due to the modification. If decreased, the new form would be less likely to establish self-sustaining populations, but it could take many generations before the genetically modified fish disappeared.

The capacit

The capacity to breed will be influenced by the ability of the fish to survive in the open environment but may also be dependent on the transgene or on the site of insertion of the gene into the genome. If the GM fish is capable of reproduction in the open environment, it may be able to spread to other habitats, establish viable populations, transfer genetic material and compete with other organisms.

The genetic modification may cause physical or physiological changes. An introduction of a growth hormone to increase growth rate may require a greater food intake, or result in the modified fish eating smaller relatives. This would deplete the stocks of 'wild-type', even if the modified fish was less fit, or even sterile. A fish modified for increased tolerance to low temperatures may have an increased capacity to scapacity to spread into a wider range of habitats. Phenotypic changes may provide the fish with a competitive advantage for food, shelter, mates and suitable breeding sites. Damaging competition with wild stocks has to be considered as a negative outcome of an introduction.

Successful breeding between modified fish and wild relatives could result in a change of the genomes of the wild stocks. Problems which might arise would then depend on the inserted gene, the frequency of transfer and the fate of the offspring. The spread of the gene from modified fish to other species by hybridisations would be of concern if the resultant hybrid offspring carried an undesirable trait.

Damage to the environment may be delayed and lo be delayed and long term as it may be caused by the descendants of the originally released fish.

Once the hazards have been identified, the likelihood of these occurring must be estimated, and risk management procedures must be considered. A serious problem is that physical containment is rarely perfect -- Penman (reference ) suggests that the introduction of modified fish into aquaculture should be considered as an intentional introduction.

It may be possible to use biological containment, by (for example) ensuring sterility.



This chapter has attempted to summarise the process of risk assessment. It has attempted to indicate that the potential impact on the environment of genetically modified organisms is not necessarily simple to identify and requires a careful analysis of all possible problems which could occur, and which need management. The chapter started with an acceptance that risk assessment implied a pessimistic tone to the use oe to the use of genetically modified organisms, but the analysis provides a method of managing problems to allow the benefits of the new technology to be expressed.


  1. Draft Compendium of Advice: Advisory Committee on Genetic Modification, Health and Safety Executive, London, April 1997.
  2. Framework for Environmental Health Risk Management: The Presidential Commission on Risk Assessment and Risk Management, Volumes I and II, Washington, 1997. See for the full text of these reports.
  3. Mackenzie DR and Henry SC (1990) "Towards a consensus" in Mackenzie DR & Henry SC (Eds) Biological Monitoring of Genetically Engineered Plants and Microbes, Proceedings of the Kiawah Island Conference Agriculture Research Institute, Bethesda, Maryland USA pp 273-283.
  4. Sustainable Development. The UK Strategy, CM2426, Her Majesty's Stationary Office (1994) ISBN 0 10 124262 X
  5. Fischoff B, Slovic P, Lichtenstein S, Read Sein S, Read S, and Combs B. (1978) How safe is safe enough? A psychometric study of attitudes towards technological risks and benefits Policy Sciences 9 127-153 and Slovic P (1987) Perception of risk Science 236 280-285.
  6. Use of Risk Assessment within Government Departments: Report prepared by the Interdepartmental Liaison Group on Risk Assessment. Health & Safety Executive, London, 1966. p3, paragraph 12.
  7. Grinstead J (1995) "Risk Assessment and Contained Use of Genetically Modified Microorganisms", in Genetically Modified Organisms: A Guide to Biosafety, Tzotzos GT (Ed) CAB International, Oxford, ISBN 0 85198 972 1
  8. Grinstead J (1995) "Risk Assessment and Contained Use of Genetically Modified Microorganisms", in Genetically Modified Organisms: A Guide to Biosafety, Tzotzos GT (Ed) CAB International, Oxford, ISBN 0 85198 972 1, p26-27.
  9. Grinstead J (1995) "Risk Assessment and Contained Use of Genetically Modified Modified Microorganisms", in Genetically Modified Organisms: A Guide to Biosafety, Tzotzos GT (Ed) CAB International, Oxford, ISBN 0 85198 972 1, p28
  10. modified organisms are generally used under conditions of 'Good Industrial Large Scale Practice' (GILSP) defined by a working group of OECD.
  11. Advisory Committee on Genetic Modification: Laboratory Containment Facilities for Genetic Manipulation ACGM/HSE Note 7 (1993) UK. These notes are currently being re-written to form a compendium of advice on the contained use of modified organisms, and should be available in the new form during 1997.
  12. Advisory Committee on Genetic Modification: Laboratory Containment Facilities for Genetic Manipulation ACGM/HSE Note 8 (June 1988) UK.
  13. Advisory Committee on Genetic Modification: Laboratory Containment Facilities for Genetic Manipulation ACGM/HSE Note 8 (June 1988) UK.
  14. Organisation of Economic Co-operation and Development Group of National Experts on Safety in 1986, ISBN 92-64-12857-3.
  15. Advisory Committee on Genetic Modification: Laboratory Containment Facilities for Genetic Manipulation ACGM/HSE Note 6 (1987) UK.
  16. Levin M: 1996 :personal communication
  17. Persley GJ, Giddings LV and Juma C (1992) Biosafety: The Safe Application of Biotechnology in Agriculture and the Environment. ISNAR, The Hague.
  18. Tiedje JM, Colwell RK, Grossman YL, Hodson RE, Lenski RE, Mack RN & Regal PJ "The Planned introduction of Genetically Engineered Organisms: Ecological Considerations and Recommendations" Ecology, 70, 298-315
  19. Biological Risk Assessment: An Editiorial Overview of some Key Policy and Implementation issues Tzotzos GT, in Genetically Modified Organisms: A Guide to Biosafety, Tzotzos, GT (editor) CAB International, Oxford 1995: ISBN 0 85198 972 1.
  20. OECD (1992) Report of the OECD Workshop on the Monitoring of Organisms introduced into the Environment. Organisation for Economic Co-operatiic Co-operation and Development, Paris.
  21. Williamson M, Perrins J, Fitter A (1990) Releasing genetically engineered plants. Present proposals and possible hazards. Trends Ecol. Evol. 5, 417-419.
  22. UK DOE/ACRE 1993: Guidance Note 1 The regulation and control of the deliberate release of genetically modified organisms. United Kingdom Department of the Environment, London.
  23. OECD Environment Monograph No 99. "Commercialistaion of Agricultural Products derived though Modern Biotechnology - Survey Results" Paris , 1995
  24. From EEC Directive 90/220
  25. Risk to human health and the environment
  26. gene transfer between species mediated through repeated back crossing of hybrid individuals to one of the parents.
  27. movement of genes between populations or between different parts of the same population.
  28. Risk in the context used here, is risk to humans and the environment, including other crop plants. see Dietz A (1993) Risk as93) Risk assessment of genetically modified plans introduced into the environment . In Transgenic Organisms, Wörmann K & Tomiu J (eds.) pp 209-227. Birkhauser Verlag, Basel, Switzerland
  29. Calgene 1990 Kanr Gene: Safety and Use in the Production of Genetically Engineered Plants, Request for Advisory Opinion. Calgene Inc. California, USA.
  30. Opinions on the Assessment of Novel Food, European Commission Scientific Committee for Food. January 1997.
  31. Penman et al (1995) "Genetically Modified Fish Populations" in Environmental Impact of Aquatic Biotechnology, OECD, pp 22 - 27. ISBN 92-64-14666-0. Papers given at a meeting organised by OECD in Trondheim, Norway in June 1993.
  32. Guidance for Experimental Release of Genetically Modified Fish DOE/ACRE Guidance Note 8, Department of the Environment UK 1997 ISBN 1 85112 024 6.