A. Charge
In July of 2001, the Food and Drug Administration’s (FDA or the Agency) Center for Veterinary Medicine (CVM or the Center) issued an open letter (CVM Update 2001) to producers of animal clones to ask them to refrain from putting edible products from those animals into the food supply until the Center evaluated the safety of those foods.1 This request had already been made to companies engaged in cloning food-producing animals during the previous year. The overall strategy chosen by the Center was to perform a risk assessment in order to determine what hazards might be introduced into animals as the result of the cloning process, to characterize the resulting potential risks, and to develop risk management proposals commensurate with the identified risks.
B. General Discussion of Risk/Safety Analyses
1. Risk and Safety
Risk and safety can be thought of as two sides of the same coin. In general, the answer to the question of “Is it safe?” is addressed scientifically by determining the conditions under which the substance or action in question is not safe, and then limiting exposures to conditions outside those limits. Because knowledge is always incomplete, and not every circumstance can be controlled, there is no such thing as “absolute safety” or “zero risk.” Risk assessors attempt to identify conditions under which risks are estimated to be as low as possible, and risk managers use that information in developing policies to protect human or animal health. The methodology used to characterize potential risk is referred to as risk assessment.2 One of the real values in performing a risk assessment is that in addition to arriving at an outcome, the process of arriving at an answer provides a framework by which data can be organized, analyzed, and interpreted. By dividing the risk assessment process into discrete steps, and then reintegrating them into an overall characterization of potential risk, risk assessment allows both the details and the “big picture” to be addressed for complex problems.
Discussion of uncertainty must accompany every risk assessment. Uncertainties may stem from a lack of fundamental understanding of biological processes and/or from data gaps that may be filled with the appropriate empirical studies; they may be exacerbated by intrinsic variability in datasets. Given that the process of risk assessment identifies data gaps and helps direct the acquisition of data that decrease uncertainties, it should not be thought of as a process that is performed just once, but rather a recursive process in the responsible development of research programs, new products, and science-based regulatory strategies.
1 http://www.fda.gov/cvm/CVM_Updates/clones.htm
2 Appendix A provides an overview of risk and safety assessments, especially as they have evolved to address issues relevant to cloning, and may be useful background reading for individuals not familiar with the processes.
2. Risk Assessment vs. Risk Management
Risk management can be defined as the set of activities of identifying and evaluating alternative strategies (often regulatory) to deal with the risks characterized in the risk assessment, and then selecting among them based on social, economic, ethical, and political conditions or criteria (NAS 1996a). Risk managers choose among different options based on the risk assessment, which is generally regarded to be relatively value-free,3; and their understanding of and responsibilities to the broader social or economic constructs within which they operate. Risk-benefit or risk-risk decisions are risk management, as they involve an active choice between two or more possible courses of action. A proposed risk management plan based on this Risk Assessment is presented in the accompanying document.
C. Risk/Safety Assessment of Cloning
In order to address the hazards and risks to animals involved in cloning and the food products derived from them (and their progeny4 ) four issues must be addressed: identifying hazards and risks; determining the degree to which existing data address questions of safety; characterizing residual uncertainties; and selecting the most appropriate risk metric for the Risk Assessment:
3Although risk assessment is based on science and relatively value-free, it generally contains a few policy-based judgments such as the selection of health protective (conservative) defaults when data are incomplete or when choosing among datasets of equal quality. The selection of policy-driven alternatives should be explicitly discussed in the risk characterization, and the implications of such choices should be described in a risk assessment.
4An animal clone is one arising directly from a somatic cell nuclear transfer event. A progeny animal is one derived from sexual reproduction that has at least one animal clone as a parent (but could result from two animal clones mating).
(1) Identifying hazards and risks. As there are no existing risk paradigms for animal clones and the food products that may be derived from them, this assessment attempts to identify hazards and risks based on the available data and consideration of the biological processes affected by cloning.
(1) Identifying hazards and risks. As there are no existing risk paradigms for animal clones and the food products that may be derived from them, this assessment attempts to identify hazards and risks based on the available data and consideration of the biological processes affected by cloning.
(2) Determining the degree to which existing data address questions of animal health or food consumption risk. At the time of this writing most of the peer-reviewed publications report on the ability to generate live animal clones from various donor cell sources and culture conditions; the frequency of successful outcomes (where success is defined as a surviving dam and a live offspring with no apparent abnormalities); and the nature and frequency of developmental errors. The nature of the published reports, with some exceptions, reflects the institutions producing them: academic laboratories tend to report the development of new technologies and the observation of abnormalities, while corporate entities tend to report successful implementation of the technology, including summaries of the health status of animal clones. These studies are useful in identifying hazards to the health of animals involved in the cloning process and characterizing the potential risks that may stem from those hazards. Despite the extensive literature search performed, and the large number of papers that were reviewed for animal health, few reports directly addressed food safety.
Reports from the peer-reviewed literature likely suffer from “publication bias,” an artifact of the criteria used to determine the “attractiveness” of publication in leading peer-reviewed journals. In general, investigators tend to submit to journals, and journals tend to publish, novel findings or hypothesis-testing results rather than surveys of the health of cohorts of animals. With a few notable exceptions, the literature on animal clones tends to consist of reports of studies of the role of various technical manipulations on the success of cloning procedures, descriptions of initial successes of cloning in species that have not yet been cloned, or descriptions of adverse outcomes. Much of the work in which cloning has been refined (and is therefore more likely to be successful) is being performed by the private sector. Given the competitive nature of the breeding and biotechnology industries, as well as the need to maintain business confidential information, much of the important information on more recent cloning outcomes has not been published or made publicly available. In order to keep the current analysis transparent to the public, however, this assessment only cites information that has been published in peer-reviewed journals, or otherwise made available to the agency by companies engaged in cloning, with explicit permission for release to the public.
(3) Characterizing residual uncertainties persisting following a review of the existing data. Due in large part to the novelty of the technology, the concentration of data at the earliest stages of clone development, and limited data directly addressing food safety, uncertainty will persist in any estimates of risk associated with animal cloning. As with all science-based uncertainties, additional data may increase the confidence with which judgments are made. The decision as to “how much is enough,” however, is a function of the nature of the risk(s) (i.e., its severity), the quality and consistency of the data (i.e., the weight of the evidence), and the tolerance of the risk management policies for uncertainty.
(4) Selecting the most appropriate risk metric for this risk assessment. The most appropriate standard to apply to the potential risk(s) associated with the consumption of foods derived from animal clones and their progeny is whether such food poses any additional risk relative to that derived from sexually-derived animals. For the purposes of this risk assessment, conventional animals are defined as those animals derived by any reproductive means other than SCNT.
D. Transgenic Animal Clones
This risk assessment addresses “just clones,” that is, animals derived via SCNT whose donor genomes have not intentionally been modified by molecular biology techniques. Transgenic clones, on the other hand, are clones whose donor cells contain exogenous heritable DNA inserted by molecular biology techniques. They are considered to occupy a different “risk space” from “just clones” because the transgenic event (the insertion of a heritable DNA sequence) is intrinsically accompanied by a series of potential risks. These include those associated with the DNA construct and those associated with the product of the gene (if there is a gene product). Organisms derived from transgenic cells will have risks specific to the inserted construct, its insertion site, and its subsequent expression. Although it is entirely possible for transgenic clones (or any transgenic organism) to be produced safely and to be a safe source of edible products, the risks associated with each animal must be determined separately on a case-by-case basis, because of the added genetic material.
Nonetheless, much of the literature on animal clones reports on experiences with transgenic clones. In some cases, the transgenic nature of the animals is explicit (e.g., Hill et al. 1999), but in many others, only careful reading or tracing back references cited in the methods section of the papers allows the reader to learn the transgenic status of the clones (i.e., Lanza et al. 2001, Cibelli et al. 2002). The question is whether any information from the transgenic clone reports can inform the identification of hazards and characterization of risks associated with “just clones.”
After a careful review of the key papers addressing transgenic clones, CVM has decided that it is not possible to determine with certainty whether any particular adverse outcome is due to the process of cloning, the transgenic nature of the donor cell, or some combination of the two. Clearly, the insertion of exogenous DNA introduces a set of hazards not present in non-transgenic clones, and by inference, the creation of a different set of risks. If transgenic animals appear to be normal, the logical inference is that neither cloning, nor transgenesis (or the combination of cloning and transgenesis) has perturbed the animals’ development. This is the case of transgenesis and cloning posing no significant (or apparent) risk. In either case, this risk assessment puts greatest weight on reports of outcomes from non-transgenic animal clones, and uses studies of transgenic clones for secondary or corroborative purposes. Nonetheless, given the large proportion of the peer-reviewed literature that reports on transgenic clones comprise, these studies have been cited with the preceding caveats. A more complete discussion of this topic is found in Appendix D: Transgenic Clones.
E. Methodology Development
When considering how to develop a risk assessment methodology for animal cloning, it became apparent the need to develop a framework that could be applied to both animal health and food consumption risks. Because the initial review of the data indicated that there were no studies explicitly evaluating the safety of food products from animal clones, the health status of the animals producing food would have to contribute to both the animal health and food safety components.
Interpreting hazard and risk from the same dataset but for different sets of receptors (animal health - the animals involved in producing clones; food safety - the consumers of the food products) requires shaping the manner in which the data are evaluated to suit the ultimate outcome of the assessment. For example, identification of adverse outcomes for animal health requires evaluating data on both surrogate dams carrying pregnancies and resulting clones. For food consumption, however, animal clones that would be condemned at slaughter, as currently practiced with conventional food animals were excluded from the analysis, an emphasis is placed on the identification of unique hazards to food consumers that could arise as the result of the cloning process. As described in the following section, and Chapter VI, this requires evaluating the dataset at a finer level of resolution than for animal health outcomes.
The net effect of the different ultimate outcomes of the animal health and food consumption risk assessments is that although the datasets considered by both assessments may overlap considerably, the manner in which they are evaluated differ, and the conclusions generated from the same (or largely overlapping) datasets vary with respect to the amount of risk present.
1. Hazard Characterization
Identifying and characterizing potential hazards is the first step in characterizing the nature of risks due to cloning (see Appendix A). CVM therefore sought to develop a framework in which adverse outcomes associated with cloning could be presented in a systematic manner that would facilitate interspecies comparisons of outcomes.
For food safety purposes, the scientific and regulatory communities have traditionally operated under the principle that domestic animals (i.e., cattle, swine, sheep, and goats) commonly consumed for food have not developed specialized organs producing toxicants to kill prey or avoid predation (e.g., venom producing glands). Further, because the components of animal tissues are necessary for life, and closely resemble the processes in humans, it is highly unlikely that “silent” pathways to produce intrinsic toxicants exist. Thus, “it is convention that animal metabolites are not considered to be natural toxicants” (Watson 1998).
In order to generate a viable clone, the differentiated genome of the donor cell or nucleus must be reprogrammed by the recipient oöplasm. Because no additional genes are being added, and the presumption is that there are no silent pathways to produce intrinsic toxicants, the only method by which hazards may arise in animal clones is from the incomplete or inappropriate reprogramming of the genetic information from the donor somatic nucleus (i.e., epigenetic effects). These phenomena are described in more detail in Chapter IV.
Where, then, would the potential hazards in clones arise? As outlined in Chapter II, during the development of an embryo, a complex series of molecular events are responsible for balancing gene expression from the maternal and paternal genomes, and directing the appropriate expression of genes in the developing embryo and mature mammal. This process is referred to as “reprogramming.” Alterations in gene expression due to those changes are referred to as “epigenetic” variability, and are present normally in conventional animals, including humans.
The most severe errors in reprogramming will result in death, obvious malformations, or metabolic derangements, and are reflected in the low “success rate” of cloning, the perinatal difficulties observed in some newborn clones, and occasional examples of altered metabolic pathways in very young animals (see Chapters V). These are clearly the subject of the animal health risk assessment. Because animals found to have a disease or condition that would render them adulterated (e.g., unfit for consumption, unhealthful, unwholesome) are prohibited from entering the human food supply, however, the only remaining food consumption hazards arising from gene dysregulation would be those that allow an animal clone to develop with apparently normal functions, but with sub-clinical physiological anomalies.
These subtle hazards are outside the conventional range of hazards commonly the subject of food safety analyses, and can be divided into three overall classes:
(1) Alterations in gene expression that lead to phenotypic variability such as coat color, size, behavior, longevity;
(2) Disruption of immune function; and
(3) Alterations in metabolism leading to changes in physiological “set-points” such that the animal has apparently compensated and appears to be normal on gross inspection, but whose physiology may be aberrant.
It is important to note, however, that changes in gene expression in individuals sharing identical genotypes have been observed in conventional animals and in humans. This phenomenon is often referred to as phenotypic variability, and can be seen at the human level in the different fingerprint and freckle patterns that identical (monozygotic) twins possess. Non-clone mice of identical genotypes fed different levels of certain nutrients can have different coat colors, and exhibit significant differences in body weight and lifespan (Cooney et al. 2002).
2. Potential Risks
Risk is defined as the probability of an adverse outcome given that exposure has occurred. This concept is often presented in the format of the “risk equation” that may be expressed as
Risk µ foutcome (exposure, hazard)
or, stated more simply, risk is some function of exposure and hazard.
The “risk equation” can be run in the forward or reverse direction. Characterizing risks from a set of hypothetical hazards is a case of running the equation in the forward direction: it allows the estimation of the probability that adverse outcomes might occur once changes that create hazards have occurred. Such approaches are useful when there is some understanding of the underlying biological processes being evaluated. For example, if incomplete genetic reprogramming (a change that may result in a hazard) were to result in animals with altered calcium transport mechanisms, a possible animal health risk could be bone weakness or malformation, and a possible food consumption risk (a probability of an adverse outcome) could be compromised human nutrition resulting from a diet of milk containing lower than expected calcium levels (the adverse outcome). In the case of the animal health risk, the degree of risk could vary from insignificant, in which no physical symptoms were present, to severe, in which the animal could experience misshapen or fragile bones leading to difficulties in walking. Because animals found to have a disease or condition that would render them adulterated (e.g., unfit for consumption, unhealthful, unwholesome), only the animals without obvious visible anomalies (and therefore less severe calcium transport anomalies) would be sources of edible products. The food consumption risks then could possibly arise from a lower available calcium pool accessible to milk production, and thus a potential nutritional risk to individuals consuming milk from such animals.
Analysis of end products such as milk constituents is an example of running the risk equation in the reverse direction: it captures the potential outcome(s) of the biological changes, and allows for the identification of exposures and hazards responsible for the risk(s). The nutritional hazard identified in the preceding example might be detected more efficiently by a compositional analysis of milk. Compositional analyses, however, are limited by available analytical methods and comparators. As far as CVM is aware, no complex food (e.g., bacon, beef steak, milk, cheese) has been fully characterized with respect to its chemical composition.5; The organisms that are or make up foods are comprised of hundreds of thousands of chemical substances that can be influenced qualitatively and quantitatively by diet, environmental conditions, and genetics. Attempts to characterize all of the chemical constituents of “milk” or “meat,” then, are neither practicable nor desirable (NAS 2004). Instead, milk and meat analyses have tended to be limited to characterizing proximates (e.g., water content, proteins, fats, carbohydrates, minerals, ash), or, when necessary or desired, to profiles of particular nutrients, anti-nutrients, or individual components of interest (e.g., vitamin content, fatty acid profiles, or protein composition).
5The International Life Science Institute (ILSI) is currently coordinating an effort to generate a database of the known chemical constituents of major food crops (e.g., corn, soy, wheat).
3. Proposed Approaches
a. Animal Health Risks
The Center determined that at this point in the development of the technology, risks to animal health are best characterized using a retrospective approach. In other words, CVM approached this issue by recording and cataloguing adverse outcomes in a biological context, rather than by elucidating specific examples of gene dysregulation and searching for their physiological sequellae. The Critical Biological Systems Approach (CBSA), described below, provides a framework in which this may be accomplished (Figure III-2). In general, the Center has relied on integrated physiological measurements to survey animal health, although it is likely that genomics, proteomics and metabolomics will see increased use for such purposes in the future. At the time that this risk assessment was prepared, however, these methods had not been sufficiently developed and validated to allow them to be used as survey tools.
b. Food Consumption Risks
Determining the safety of food products from animal clones and their progeny, at least in its earliest stages, is likely best accomplished by using both approaches: prospectively drawing on knowledge of biological systems in development and maturation, and in retrograde, from an analysis of food products. An intrinsic and valuable part of this analysis is cataloging the available information, and identifying data gaps and uncertainties that may in turn suggest research that could serve to decrease the identified uncertainties. The following sections describe the methodology CVM has proposed to accomplish a rigorous, science-based analysis of potential hazards and risks associated with the consumption of food products derived from animal clones and their progeny.
Prior to undertaking such an analysis, however, subtle hazards and potential risks that may be posed by animal clones must be considered in the context of other mutations and epigenetic changes that occur in all food animal populations. Some are considered beneficial, and have been selected for by animal breeders when a desirable phenotype is obtained. For example, not-so-subtle genetic mutations that have occurred at least twice in nature are the development of double-muscled beef breeds such as the Belgian Blue and Piedmontese, which arose from different mutations in the myostatin gene (McPherron and Lee 1997). These animals appear to be healthy, although sexual maturity appears to be delayed relative to other breeds, and female fertility appears to be somewhat lower. Nonetheless, these animals are used in selective beef breeding programs in several countries as they have 20-30 percent more muscle mass than cattle with the wild-type myostatin gene, feed efficiency is increased, and the meat is considered to be more tender although lower in fat content. Meat from these animals is presumed to be food, and as such enters the food supply with no additional regulatory scrutiny. Epigenetic changes that occur on a regular basis include variations in pigmentation patterns (e.g., coloration patterns on Holsteins) and are perhaps most easily thought of as those differences observed in identical twins, such as different fingerprints and freckle patterns.
Finally, it is important to remember that any discussion of subtle hazards and potential risks associated with the products of animal clones is not conducted in a regulatory vacuum. All food, including that from animal clones, must meet existing regulatory requirements in order to be marketed lawfully in the United States.
F. Two-Pronged Approach to Assessing Food Consumption Risks
Given the assumption that food derived from clones will be in compliance with existing regulatory requirements for food products from conventional animals, CVM proposes a two-pronged approach for evaluating the potential risks associated with the food products of animal clones and their progeny (AC/P) (Figure III-1). The first component, the Critical Biological Systems Approach (CBSA) is based on the hypothesis that a healthy animal is likely to produce safe food products, and incorporates a systematic review of the health of the animal clone or its progeny. The second component, or the Compositional Analysis Method, is based on the operating hypothesis that food products from healthy animal clones and their progeny that are not materially different from corresponding products from conventional animals are as safe to consume as their conventional counterparts. It relies on the comparison of individual components of edible products, and the identification of the appropriate comparators.
Figure III-1:
[ D ]
1. Critical Biological Systems Approach
a. Overview
The CBSA (Figure III-2) is based on a cumulative evaluation of health status indicators of animal clones. Mechanistically derived, it considers SCNT and the subsequent development of the animal clone from a biological “systems analysis” perspective, and thus may be thought of as being “HACCP6 -like.” It accepts that at this time SCNT is a biologically imprecise and inefficient process resulting in few live births relative to the number of implanted embryos, and that some animals are born with obvious defects or subtle anomalies. It also assumes that biological systems are capable of repair or correction, either intrinsically or following human intervention. For example, animals that may have difficulty surviving on their own immediately after birth may develop into healthy, reproducing individuals if provided support in the form of respiratory assistance and warmth during the period immediately after birth. Alternatively, these animals may not recover, and may remain “sickly” or unthrifty until they are culled.
The cumulative nature of the CBSA allows for the incorporation of both favorable and unfavorable outcomes. The former, provided that all other measures appear to be normal, will result in the judgment that the animal will produce food that is safe for consumption; the latter implies that animals with anomalies may be unsuitable for food.
6HACCP is the Hazard Analysis Critical Control Point approach adopted by USDA and FDA for assuring the safety of certain food products undergoing some degree of processing. http://www.fsis.usda.gov/Science/HACCP_Models/index.asp
Figure III-2: Critical Biological Systems Approach
[ D ]
b. Evaluation Nodes
The CBSA selects five key developmental stages of an animal clone’s life, analogous to the “critical control points” of the HACCP analysis. These stages provide biologically-based developmental “collection nodes” (Developmental Nodes) (indicated in
Figure III-2 by periwinkle-colored boxes) that also serve as agronomically appropriate points at which to collect data. Examples of the types of data that could be collected are illustrated in Figure III-2 as yellow boxes. It is important to note that these Developmental Nodes address functionality and not necessarily discrete time points, as the latter will vary among species and breeds.
Developmental Node 1 incorporates the initial technical steps involved in SCNT, including cell fusion through implantation, and subsequent embryo and fetal development. Chapter IV covers many of the early common molecular events that occur during this time period common to mammals; Chapter V reviews these issues as they impact on the health of clones and their surrogate dams; and Chapter VI reviews these steps from the perspective of identifying food consumption risks.
Developmental Node 2 encompasses the Perinatal period, including late gestation, labor induction in the dam, delivery, and the critical time period of approximately 0-72 hours after birth. This developmental node allows for the analysis of animal health data relevant to both the surrogate dam and the clone, although few food consumption risks are anticipated to occur at Developmental Node 2 because clones of that age would not be consumed as food.
The third developmental node (Developmental Node 3), Juvenile Development and Function, encompasses the period of rapid growth between birth and the onset of puberty, and may vary in duration among the species considered.
The Reproductive Development and Function Node (Developmental Node 4) encompasses puberty and reproductive function throughout the reproductive period of the animal. Food consumption risks arising from milk production may first be encountered at this point of the animal’s life. Because of the complex integration events that must occur for effective reproduction to take place, this developmental node is critically important for evaluating the health and functionality of animal clones. Proper reproductive function indicates that the complex and inter-related physiological systems required for the development and delivery of functional germ cells (and, in the case of females, viable offspring) have occurred appropriately.
The Post-Pubertal Maturation Node (Developmental Node 5) encompasses all non-reproductive functions of sexually maturing or mature animals, including growth, weight gain, disease frequency, aging, and lifespan, where available.
Because the value of clones lies in their use as breeding stock (and is reflected in their relatively high cost), “founder” animal clones are not likely to be slaughtered initially for meat. It is anticipated that most of the food products, especially meat, from clone lineages will enter the food chain as the progeny of animal clones, or their subsequent offspring. Milk from dairy clones could enter the food supply, following breeding and delivery of offspring. Meat from clones could enter the food supply if, for instance, conditions outside the producer’s control forced herd culling (e.g., loss of funding), or when older animals reach the end of their functional utility (e.g., loss of fertility in breeders). Table III-1 summarizes the Developmental Nodes, the types of data likely to be collected at each node, and the potential for the entry of clones into the food supply.
|
Table III-1: Summary of Developmental Nodes and Implications for Food Consumption Risks | ||
| Developmental Node | Types of Observations/Data | Potential for Entry into Food Supply |
| 1: Cell fusion through implantation, embryo and fetal development | Selection of donor and recipient cells, oöcyte maturation and activation, fusion method, days in culture, culture conditions, number of fusions, number of blastocysts formed (if measured). Number of implantations, early and late gestation losses, placentation, pregnancy maintenance, morphological anomalies. | None |
|
2: Perinatal period including immediate pre-partum, delivery, and up to 72 hours post-partum |
Number of animals delivered with/without assistance, survival, morphological abnormalities, post-parturition survival, physiological/biochemical characterizations of surviving/dead animals. | Minimal, due to low likelihood of entry into food supply as meat, except for injured animals. |
| 3: Juvenile Development (cattle: pre-weaning; swine, sheep, goats: post-weaning period) |
Survival rate, measures of growth, physiological and biochemical markers of health status. | Relatively low, but possibly as meat (e.g., veal, lamb, suckling pig). |
| 4: Reproductive Development and Function | Development of secondary sex characteristics, spermatogenesis, oögenesis, gender appropriate behavior, age of pubertal onset. Fertility measures for males and females. For females, mothering behavior, milk production. | Milk |
| 5: Post-Pubertal Maturation | Growth, weight gain, muscle/fat ratios, milk production. | Meat, Milk |
G. Limitations of the Risk Assessment
This is a qualitative, comparative risk assessment that does not attempt to assign a quantitative value to estimates of risk or safety. The strongest conclusions that can be drawn regarding positive outcomes in risk assessments of this type are “no additional risk” because outcomes are weighed against known comparators. If a finding of “no additional risk” were to be applied to the health of animal clones, it would mean that the cloning process would not pose any greater risk to the health of the animals involved than other ARTs. Applied to the safety of edible products derived from clones, a finding of “no additional risk” would mean that food products derived from animal clones or their progeny would not pose any additional risk relative to corresponding products from non-clones, or that they are as safe as foods that we eat every day. As with all risk assessments, some uncertainty is inherent either in the approach we have used or in the data themselves (for a more complete discussion of the uncertainties in this Risk Assessment, see Chapter VII).
|
||||||
 |
||||||
|
||||||