TEAM APPROACH AND SHARED RESPONSIBILITIES
A team approach to animal-use protocol development and animal management is valuable for meeting research objectives while maximizing attention to animal care. The team approach relies on the idea of shared responsibilities. Participants include neuroscientists and their laboratory personnel; veterinarians; animal-husbandry staff; the IACUC chair, members, and staff; and the research institution.
The institution must establish a culture of respect for the animals and maintain a commitment to following the Guide, PHS Policy, and the AWRs. However, the Guide, PHS Policy, and the AWRs are not intended to be barriers to research. Working as a team, the principal investigator (PI), the IACUC, and the veterinarian should be able to devise a means of accomplishing the research goal while addressing the needs of the animals. The “performance standard” encouraged by the Guide (p. 3) is a powerful tool for IACUCs and veterinarians to use in developing strategies to promote both animal well-being and good science.
Developing an animal care and use protocol is a negotiation among the PI, the veterinarian, and the IACUC to balance animal well-being with the experimental goals. By involving all parties early in the protocol-development process, particularly for protocols that involve extensive experimental manipulation or difficult to maintain animal models, the PI can help to prevent misunderstandings and delays while facilitating IACUC review and approval.
Development of Protocols
The PI and research staff develop the scientific concept that underlies an animal protocol. The PI is the expert in the scientific goals and the experimental design and is often highly knowledgeable about animal use and well-being. The veterinarian is the trained expert in the latter subject and can advise the PI and the IACUC on performance-standard development and implementation. However, the IACUC has the final responsibility and authority for evaluating the protocol outcome and approving exceptions to guidelines and regulations if any are necessary. The veterinarian should be involved in animal-use protocol development, preferably before the protocol is submitted to the committee for official review. Indeed, according to the AWRs, the veterinarian must be consulted for any procedure that may cause more than momentary or slight pain or distress (AWR 2.31(d)(1)(iv)(B)). In some cases, veterinary input is gathered primarily during the final protocol-review process, and this delay can complicate protocol review and prolong the approval process.
The IACUC staff and/or chair can solicit additional information and perform a prereview of the animal-use protocol to help the PI recognize portions of the protocol that require clarification or additional information. The IACUC must consider sample size, pain and/or distress, and experimental and humane endpoints among other considerations. The IACUC may request direct observation of a procedure (particularly a new or unusual procedure), pilot studies, and particular experimental measurements or monitoring procedure to evaluate and ensure animal well-being.
Execution of Protocols
The research team has the primary responsibility for animal assessment and intervention. However, researchers and the animal-care staff must coordinate their efforts to provide appropriate animal care and monitoring. Unanticipated adverse effects of the research that are or may be a threat to the health or safety of the animal must be reported to the IACUC immediately. As an animal-use protocol must describe any anticipated adverse effects, if an unanticipated adverse effect that was or could be a threat to the health or safety of the animal were to occur, then the protocol does not accurately reflect the animal-use activity and must be modified accordingly and reapproved.
The veterinary staff has the legal responsibility for animal care. Veterinary medical care is best administered with consideration for the scientific goals of the study. However, the veterinarian must have institutional authority to make decisions on behalf of the animal in critical situations.
The husbandry staff has day-to-day responsibility for assessment of animal well-being, regardless of experimental use. The caretaker staff is in a unique position to observe large numbers of animals and to understand the
animals both individually and as a species or strain. They observe eating, drinking, urination, defecation, behavioral abnormalities, and subtle signs of problems. Because they observe large numbers of normal animals, the animal-care staff can be key participants in determining the phenotype of knockout and transgenic rodents.
The IACUC, after review and approval of the animal care and use protocol, has the responsibility to ensure that procedures are carried out in accordance with the protocol. The IACUC may request periodic reports from the PI or from the veterinarian monitoring the research. During semiannual facility inspection, the IACUC can verify that procedures are consistent with the approved protocol.
The institution provides the animal-research infrastructure in the form of core facilities for animal care, mandates and resources for training, and the occupational health and safety program. The institutional official (IO) is the individual who is authorized to legally commit on behalf of the research facility that the requirements of 9 CFR parts 1, 2, and 3 will be met (AWR 1.1). The IO is generally the senior administrator with authority to commit institutional resources to ensure compliance with governing regulations and guidelines. The IO should empower the appropriate people to intervene on behalf of the animals, provide adequate facilities and staff, and take the lead in creating a compliant and responsible institutional culture.
Pilot studies are integral elements of animal experimentation. As stated in the Guide (p. 10), “if little is known regarding a specific procedure, limited pilot studies designed to assess the effects of the procedure on animals, conducted under IACUC oversight, might be appropriate.” Numbers of animals in pilot studies are usually small and the researcher and veterinary staff should closely monitor these special kinds of projects. The novelty or unpredictable nature of the experimental techniques or animals being investigated in pilot studies warrants heightened awareness of animal care and welfare.
An important goal for pilot studies should be the collection of the maximal amount of useful preliminary information, and a team effort and team approach will be key factors in reaching the goal. The research and animal-care staff should be aware of potential concerns or complications that may arise during the pilot study. Pilot studies may pose new challenges for all, but the close interactions between scientific and veterinary staff that develop may be the cornerstones for a successful outcome of both the pilot study and the eventual research project.
Examples of situations that may warrant pilot protocols are:
The need to develop a new technique.
The need to adapt a technique that has not been used previously in a particular species.
The need to implement a procedure that uses a technique unfamiliar to the institution except indirectly through published information.
The need to modify a procedure to simplify the measurement of a variable or to improve the statistical significance or power of an experiment so as to reduce the number of animals required.
The need to refine an existing technique before adapting an approved protocol to use the refinement (testing the results of the refinement may improve the outcome for both the animals and the experimenter.)
As stated in the US Government Principles (IRAC, 1985), investigators should use the minimum number of animals required to obtain valid results (see also the Guide, p. 10, and AWR 2.31 (e)(2)). However, investigators frequently err on the side of using too few animals rather than too many (Dell et al., 2002). That results in a study that has too little power to detect a meaningful or biologically significant result. For example, in a meta-analysis of 44 animal experiments involving fluid resuscitation, Roberts and colleagues (2002) found that none had sufficient power to reliably detect a halving of death rate. To avoid this error, researchers should calculate the sample size necessary to detect a statistically significant effect. Several factors must be known or estimated to calculate sample size (Dell et al, 2002):
the size of the effect under study (difference between experimental groups)
the population standard deviation of the effect
the desired power of the experiment to detect the effect (usually 80-90%)
the significance level (usually .05 or .01).
Methods for computing sample size are found in Appendix A. In general, the smaller the effect size or the larger the population variability, the larger the sample size must be to detect a difference. It should be noted that using a more sophisticated experimental design and statistical analysis provides more power to detect an effect (Dell et al., 2002).
Some aspects of neuroscience research (such as developing and producing genetically modified animals) pose particular difficulties in estimating the number of animals necessary for a given experiment. Additional guidance on these issues is provided in Chapter 3, “Genetically Modified Animals,” and also in Appendix B.
PAIN AND DISTRESS
Pain may be inherent in the study of pain and/or distress, but it can also be an unintended aspect of the research (for example, in animal models of disease, as a byproduct of a survival surgical procedure, or in transgenic animals with a
clinical phenotype). It is critical to recognize and manage animal pain and distress.
The International Association for the Study of Pain has defined pain in humans as an “unpleasant sensory and emotional experience associated with actual or potential tissue damage, or described in terms of such damage” (Mersky, 1979). Although animals cannot communicate verbally, they exhibit motor behaviors and physiologic responses similar to those of humans in response to pain. Those behaviors may include simple withdrawal reflexes; complex, unlearned behaviors such as vocalization and escape; and learned behaviors such as pressing a bar to avoid further exposure to noxious stimulation. However, there are species-specific behaviors that animals may express in response to pain (Bolles, 1970), see Table 2-1 for review.
Stress (or the stress response) has been defined as “the biological response an animal exhibits in an attempt to cope with threats to its homeostasis” (Stokes, 2000). Threats to homeostasis are called “stressors.” Stressors can be physical, environmental, or psychologic in origin (NRC, 1992), and adaptation can involve immunologic, metabolic, autonomic, neuroendocrine, and behavioral changes (Moberg and Mench, 2000), but the type, pattern, and extent of the changes depend on the stressor involved. When the animal responds to a stressor in an adaptive way, the animal returns to a state of comfort. It is also possible for stressors to induce responses that have beneficial effects (Breazile, 1987). Animals (and people) are normally exposed regularly to stressors to which they need to respond and adapt (Sapolsky, 1998), and some stress is probably necessary for well-being (NRC, 1992).
When an animal is unable to completely adapt to a stressor and the resulting stress, an aversive state has developed defined as distress. The term distress encompasses the negative psychologic states that are sometimes associated with exposure to stressors, including fear, pain, malaise, anxiety, frustration, depression, and boredom. These can manifest as maladaptive behaviors, such as abnormal feeding or aggression, or pathologic conditions that are not evident in behavior, such as hypertension and immunosuppression (NRC, 1992).
Extensive guidelines, policies, and regulations govern the management of pain and distress in laboratory animals. The US Government Principles (IRAC, 1985) state:
Proper use of animals, including the avoidance or minimization of discomfort, distress, and pain when consistent with sound scientific practices, is imperative. Unless the contrary is established, investigators should consider that procedures that cause pain or distress in human beings may cause pain or distress in other animals [Principle IV].
Procedures with animals that may cause more than momentary or slight pain or distress should be performed with appropriate sedation, analgesia, or anesthesia. Surgical or other painful procedures should not be performed on unanesthetized animals paralyzed by chemical agents [Principle V].
The AWRs, PHS Policy, and the Guide require the IACUC to ensure that animal-use protocols include strategies for minimizing pain and distress in animals. Specifically, USDA (through the AWR 2.31 (d)(ii) and (e) and APHIS/AC Policy 12) requires the investigator to consider alternatives to procedures that
TABLE 2-1 Indicators of Pain in Several Common Laboratory Animalsa
may cause more than momentary or slight pain or distress and to provide a written narrative description of the methods and sources used to determine that alternatives to the procedure were not available. As noted in the Institutional Animal Care and Use Committee Guidebook (ARENA-OLAW, 2002), the animal-use protocol must provide sufficient information for the IACUC to evaluate the pain and/or distress potentially resulting from the study and the appropriateness of the methods proposed to minimize it. The attending veterinarian is an important team member, working with the researcher and the IACUC to ensure animal welfare when there is the potential for pain and/or distress.
Assessment of Pain
According to the Guide, “fundamental to the relief of pain in animals is the ability to recognize its clinical signs in specific species” (p. 64). Pain can be assessed by evaluating behavioral measures such as eating, socializing, and withdrawal reflexes, and physiologic measures such as heart rate and respiration rate (see Table 2-1). However, species, and even strains and individuals of the same species, may vary widely in their perception of and response to pain (NRC, 1992; Wixon, 1999). Even for an individual animal, pain sensitivity varies among different tissues and organs (Baumans et al., 1994), and pain sensitivities can be altered by pathologic processes or experimental procedures (Carstens and Moberg, 2000). For example, during the initial phase of lipopolysaccharide-induced fever, rats exhibit hyperalgesia, whereas they exhibit hypoalgesia during the later stages of the illness (Carstens and Moberg, 2000). The existence of these differences underscores the point that pain and distress exist as a continuum of experience. In addition, some animals may hide signs of pain; for example, it has been suggested that rats may mask pain during the dark-cycle hours to avoid displaying abnormal activity and increasing their risk of predation (Roughan and Flecknell, 2000).
AALAS (AALAS, 2000) suggests that the magnitude of the pain that the animal is expected to experience be categorized in the protocol and monitored and that there be an opportunity to adjust the pain category once the study is under way. It is important that researchers and animal-care staff have a solid knowledge of the normal and abnormal physiology, behavior, and appearance of the animals in their care (Anil et al., 2002; NRC, 1992).
Acceptable levels of noxious stimulation are those that are well tolerated and do not result in maladaptive behaviors. Acceptable levels range from an animal’s pain threshold to its pain tolerance level. Pain threshold is the stimulus level at which pain is first perceived, while pain tolerance is the highest intensity of painful stimulation that an animal will voluntarily accept. As the intensity of a stimulus approaches the pain tolerance level, an animal’s behavior will become dominated by attempts to avoid or escape the stimulus, and this degree of pain must be alleviated (Dubner, 1987).
It is important to note that it is usually incorrect to infer that an animal’s pain tolerance level is signaled by the onset of avoidance or escape behavior, as some avoidance-escape behavior is an appropriate adaptive response. It is only when the animal’s behavior is dominated by avoidance-escape attempts that the behavior becomes maladaptive, signaling unacceptable levels of pain (NRC, 1992).
In pain studies, giving animals control over the source of pain by allowing them to withdraw from a painful stimulus is an effective way to minimize pain and the distress associated with it. If an animal is denied control of the stimulus and it approaches the tolerance limit, maladaptive behaviors will appear, and the animal should be presumed to be in distress. Maladaptive behaviors include persistent attacks on the perceived source of the pain, self-mutilation at the injured or stimulated site, and a state of learned helplessness in which the animal gives up and no longer attempts to escape, avoid, or control the stimulus. To avoid the development of maladaptive behaviors and to minimize pain during experimental manipulations where the animal is denied control of the stimulus, it is critical that the neuroscientist attempt to define the level of pain produced by the stimulus (Dubner, 1987), and ensure that the level imposed by the stimulus is below that which causes the development of maladaptive behaviors. In most cases, previous experimental or published data will indicate the level of pain produced by the stimulus; lacking this information, a pilot study to identify the level of stimulus that produces maladaptive behavior could be useful.
Pain assessment will vary with the pain scale or scoring system used. Scoring systems involve assigning a numeric score to constellations of behavioral, physical, and physiologic observations, and this process can be subjective. There are no generally accepted objective criteria for assessing the degree of pain that an animal is experiencing, and different species or strains can vary in their response to pain. Physiologic measures include heart rate, blood pressure, and respiration rate, but obtaining most of the measures requires some degree of intervention, which may not be feasible or desirable (Baumans et al., 1994).
Recent studies on pain in animals include methods for quantifying specific motor behaviors as indirect measures of responses to mechanical, thermal, or chemical injury (Dubner and Ren, 1999). Animals will withdraw an injured body part from a stimulus, where different levels of stimulation affect the latency or force of withdrawal. This withdrawal response is considered a measure of pain, which correlates highly with more integrative nocifensive behaviors (behaviors in response to pain), such as licking of the injured body part and guarding behavior.
Some behavioral signs are usually associated with pain (Soma, 1987). Animals often communicate through posture. They may exhibit guarding behavior in an attempt to protect the injured part. Vocalizations are important indicators of pain in several species (Anil et al., 2002). Animals in pain may lick, bite, scratch, shake, or rub the site of injury. Restlessness may also be observed, including pacing, lying down and getting up, and shifting weight. Animals in pain may stay in one place for abnormal lengths of time and be reluctant to move or rise. They
may withdraw from contact with other animals. They may become listless and refuse to eat or reduce their eating and drinking. They may avoid being handled. These are all possible signs of pain, but none alone is sufficient to determine the presence or level of pain. For example, many animals vocalize intensely when they are handled even if they are not in pain (e.g., Stafleu et al., 1992). Multiple criteria should therefore be assessed (Bayne, 2000; Wallace et al., 1990).
An important step in determining that an animal is in pain is recognition of a departure from normal behavior and appearance (Dubner, 1987; Kitchen et al., 1987; Morton and Griffiths, 1985; NRC, 1992). But as Bayne (2000) indicates, assessments vary with the scale used, and the scales can be very subjective. Flecknell and Silverman (2000) noted that preprocedural scoring is necessary to obtain an appropriate baseline so that confounding variables (such as behavioral effects produced by analgesics) can be identified. For example, some of the consequences of surgery in rats, such as loss of body weight and reduction in food and water intake (signs frequently interpreted to indicate pain or distress), can also be produced in normal, unoperated-on rats by administering opioid analgesics.
Recent evidence indicates that some signs of pain may not be perceived by personnel, such as the ultrasonic vocalizations of infant mice (Nastiti et al., 1991), but are detectable with appropriate equipment. Several excellent references discuss species-specific behaviors that are indicative of pain (Carstens and Moberg, 2000; Hawkins, 2002; Morton and Griffiths, 1985; NRC, 1992; Roughan and Flecknell, in press; Soma, 1987; Wallace et al., 1990).
Assessment of pain should not be influenced by the biases of the observer (Sanford et al., 1986), and the observer should be well trained in both normal and abnormal behaviors of the species in question. Variability among observers can have a substantial effect on the interpretation of assessment data (Holton et al., 1998).
Chronic or persistent pain differs from acute pain because it may not be associated with any obvious pathologic condition and does not serve any protective function. Signs of chronic pain can be subtle and difficult to detect in that an animal’s behavior may change slowly and incrementally. Chronic or persistent pain is also more likely to lead to distress and maladaptive behavior. Signs of chronic or persistent pain include decrease in appetite, weight loss, reduction in activity, sleep loss, irritability, and decrease in mating behavior and reproductive performance (Soma, 1987). Alterations in urinary and bowel activities and lack of grooming are often associated with persistent pain. Severe chronic pain can reduce body temperature, cause a weak and shallow pulse, and depress respiration. As noted above, animals cannot control chronic or persistent pain, and it is important to assess the intensity of the pain by using behavioral measures.
Assessment of Distress
Some of the husbandry and experimental procedures that animals experience in neuroscience and behavioral research have the potential to cause distress. Distress can be either transient or prolonged and can range from mild to severe. Determining when stress becomes distress, and thus an animal-welfare concern that requires amelioration, is difficult. Our understanding of the relationship between the (measurable) physiologic changes that occur during an acute stress response and ensuing adverse psychologic states is generally poor. An animal’s mental state can be inferred only indirectly, and many factors can influence whether an animal responds in an adaptive or maladaptive fashion to a particular stressor. Those factors include genetic predisposition, experience, age, sex, species, and the social context in which the stressor occurs.
It is even more difficult to assess the effects of chronic and intermittent stressors that are likely to be experienced by animals as part of routine husbandry and housing in the research setting. There is evidence that animals can successfully adapt physiologically to most stressors experienced as part of routine husbandry and housing (Line et al., 1989; Sharp et al., 2002a,b); however, animals may not be able to physiologically adapt to all such stressors, and their ability to adapt is likely dependent on the aversiveness, duration, and frequency of the stressor (Line et al., 1989). Yet, even when animals can adapt physiologically to such chronic stressors, they may not exhibit behavioral adaptation (Ladewig, 2000). In some cases of chronic stress (for example, in severe depression), the physiologic stress system may eventually stop responding normally to challenges, even after the source of the chronic stress is removed (Sapolsky, 1998).
Nevertheless, it is imperative to evaluate and (where possible) ameliorate distress in the research environment (NRC, 2000). Several general schemes have been proposed for recognizing distress, including a measurable shift in biologic resources (Moberg, 1999), such as a change in metabolic function, or evidence of maladaptive behavior (NRC, 1992). The physiologic and behavioral changes that accompany some states of distress have been fairly well characterized. For example, common manifestations of fear or anxiety are motor tension (shakiness and jumpiness), hyperactivity of the sympathetic nervous system (sweating, increased respiration and heart rate, and frequent urination), and hyperattentiveness (increased vigilance and scanning) (Rowan, 1988).
One problem in assessing stress and distress has been that measurement techniques that involve handling, blood sampling, or tissue collection may themselves be stressors and cause physiologic changes. However, many noninvasive or less invasive methods for physiologic monitoring can now be used in the research setting. They include implanted radio transmitters to measure autonomic nervous function, microdialysis techniques for sample collection, remote blood sampling methods, biosensors for recording central nervous system responses in
freely moving animals, and measurement of hormones in hair, feces, and urine (Cook et al., 2000; Goode and Klein, 2002; Koren et al., 2002).
One of the most important noninvasive methods for assessing distress is observation of animal behavior. At least when measured in situations where animals have behavioral flexibility and choice, behaviors can provide information about what animals prefer or avoid, and hence are indicators of emotional states (Mench, 1998). Behavior is a part of an animal’s adaptive repertoire for responding to stressors, so it is important to distinguish adaptive behaviors from maladaptive behaviors, such as self-mutilation, unresponsiveness to important signals, hyperactivity, and excessive response to stimulation. Useful observation requires knowledge of the natural history and perceptual capacities of the particular species or strain of animal (Bayne, 1996), the usual frequencies and intensities of particular behaviors, and the causes and functions of the behaviors (Mench, 1998; Rushen, 2000). Because of individual variability, a baseline behavioral profile of an animal should be established if changes in behavior are going to be used to monitor the animal for distress. As with the assessment of pain, personnel assessing behavior should be knowledgeable and skilled in the interpretation of behavior, and assessments should not be influenced by the observers’ biases (Bayne, 2000).
Alleviation of Pain and/or Distress
Four general approaches are available to minimize pain (Dubner, 1987): the use of general anesthesia, the use of local anesthesia and/or analgesia, the training of animals to avoid situations that produce pain (escape-avoidance behavior), and control of the intensity and/or duration of the stimulus by the neuroscientist.
The selection of a general anesthetic should reflect professional judgment as to which anesthetic best meets the clinical and humane requirements without compromising the scientific aspects of the research protocol (NRC, 1996). That sufficient anesthesia has been provided can be ensured by monitoring reflexive responses to painful stimuli, respiration, pupil size, stability of heart rate and blood pressure or electroencephalographic activity. Occasionally in neuroscience research, surgical lesions are created producing a functional decerebration and thereby eliminating all possibility of pain and the need for general anesthesia.
The use of local anesthesia and/or analgesia is also a widely used technique for alleviating pain and/or distress. Vainio et al. (2002) provide a useful description of the clinical efficacy and adverse effects of opioids and nonsteroidal anti-inflammatory drugs (NSAIDS) for analgesia in laboratory animals, as well as the dose, route, and frequency of administration of the common opioids and NSAIDS.
For a discussion of the training of animals to avoid painful stimulation and of how investigators can control the intensity or duration of painful stimulation, see Chapter 4. In some cases, euthanasia may be the most appropriate means of alleviating pain (see Chapter 2—“Humane Endpoints”).
Distress is usually an undesirable outcome of an experiment, so strategies for avoiding or minimizing it should be identified during the planning of the study. When distress arises from unintended sources, identification and elimination of the cause is the most obvious course of action. Distress can also be alleviated pharmacologically and nonpharmacologically. Two of the major causes of distress are lack of predictability and control of stimuli (Sapolsky, 1998), and it can be useful to condition animals to experimental or husbandry procedures that they will experience or to allow them to have some control over the stimuli imposed, for example, by providing a way to escape from staged aggressive encounters (as discussed in Chapter 7) or by providing nesting material so that a thermally challenged animal can better control its body temperature. Environmental modifications that can make an animal more comfortable, including changing the ambient temperature, increasing ease of access to food and water, and determining whether social contact would ameliorate or accentuate the stress burden. Appropriate enrichment of the social and structural environment can decrease distress by decreasing boredom and fearfulness, and facilitating coping (Bayne et al., 2002; Carlstead and Shepherdson, 2000; Mench, 1998). If possible, pain-related distress should be managed pharmacologically. However, if pain is the object of the study, pharmacologic options for reducing distress may be few or unavailable. In some studies, the use of sedatives, anxiolytics, dissociative anesthetics, and/or analgesics will not conflict with study goals; IACUCs should require that these options be discussed in animal-use protocols for neuroscience or behavioral research.
In all cases the veterinarian should be consulted regarding methods to minimize pain and distress. Accepted best practices for managing unrelieved pain and distress should be incorporated into the protocol design unless there is a scientific reason to do otherwise. The neuroscientist must provide assurance that unrelieved pain or distress will not continue past the point necessary to achieve the scientific goals of the study (ARENA-OLAW, 2002). Additionally, a mechanism for prompt reporting (for example, to the veterinarian or the IACUC) of animals that have been unexpectedly distressed or pained by the study should be developed and implemented (Bayne, 2000) and should be inherent in the animal-use protocol design.
USING ANIMAL BEHAVIOR TO MONITOR ANIMAL HEALTH
Animal behavior can be an excellent measure for assessing overall health, indeed, the clinical signs used to diagnose disease in animals are often based on behavior (for example, signs of pain) (Fox, 1968)—although this approach has not been well documented in the veterinary or behavioral literature. A sound understanding of animal behavior is key for the veterinarian or other professional in assessing animal health. Recognition of the importance of behavior as related to animal health, and correspondingly to the veterinary profession, was formalized by
the American Veterinary Medical Association in 1993 when the American College of Veterinary Behaviorists (ACVB) was given specialty-board status.
In the research environment, routine behavioral observations can aid in the detection of disease in animals that are not exhibiting any other clinical signs. For example, a cynomolgus monkey was diagnosed with diabetes mellitus through initial observations of uriposia (urine-drinking) (Levanduski et al., 1992); the tentative diagnosis was then confirmed with urinalysis and blood-chemistry evaluation. Sensitive indicators of animal health include measures of food or fluid intake and performance of specific tasks (NIH, 2002).
Because of the long-term use of individual animals in a neuroscience or behavioral study, the physical proximity between researcher and animal, and the wide variety of behavioral data collected during a study, neuroscientists have an excellent opportunity to monitor animal behavior and health. Subtle changes detected in the animal’s demeanor or its willingness to work in a study or sudden changes in performance on behavioral tasks may be the first indicators of a health problem that should be investigated. If such changes are noted, the researcher should promptly notify the veterinarian so that the animal can be more fully evaluated.
Endpoints are established for both experimental and humane reasons. An experimental endpoint is chosen to mark the planned end of an experimental manipulation and associated data gathering. A contingent experimental endpoint may also be used to signal euthanasia to remove an animal from the study for humane reasons. On the other hand, in experiments with unrelieved or unanticipated pain/or distress, humane endpoints are criteria that indicate or predict pain, distress, or death and are used as signals to end a study early to avoid or terminate pain and/or distress. Ideal endpoints are those that can be used to end a study before the onset of pain and/or distress, without jeopardizing the study’s objectives. However, in most cases, humane endpoints are developed and used to reduce the severity and duration of pain and/or distress (Stokes, 2000).
Humane endpoints should reflect actual or imminent deterioration of an animal’s condition, and they should be easy to assess over the course of the study (Toth, 2000). General categories of endpoints include biologic markers, such as the development of paralysis in models of neural tumors (Huang et al., 1993, 1995); markers of therapeutic failure, such as persistent signs of tumor growth despite drug intervention; markers of disability, such as inability to stand in models of bacterial endotoxemia (Krarup et al., 1999); markers of disease exacerbation, such as increased seizure frequency; and general markers of clinical deterioration, such as substantial changes in body weight, alertness, respiration, and body temperature (Redgate et al., 1991; Toth, 1997).
Humane endpoints should be specific to an experimental model or animal strain (Toth, 2000). For example, a decrease in body temperature from 35 to 28°C has been found to be an early predictor of eventual death in studies of bacterial and viral infections, toxicoses, and activity-induced stress in mice (Gordon et al., 1990, 1998; Morrow et al., 1997; Soothill et al., 1992; Toth, 2000; Wong et al., 1997).
The first step in developing a humane endpoint is to describe the clinical progression that a particular animal or group of animals is likely to experience as a result of experimental manipulation or spontaneously occurring disease during their lifetime. Next, potential humane endpoints should be identified. One strategy for identifying humane endpoints is to closely monitor a few animals undergoing a new procedure using score sheets to record clinical, behavioral, and biochemical signs found during the progression of the experiment (Morton, 2000). Finally, a humane endpoint(s) should be selected based on its ability to accurately and reproducibly predict or indicate pain and/or distress, imminent deterioration, or death. The humane endpoint(s) must also be specific to the study (Toth, 2000). For example, the humane endpoint selected for a study of preventative treatments could allow an animal to be euthanized earlier than a humane endpoint selected for a study of disease treatments. In the first instance, the onset of disease symptoms could be a humane endpoint without jeopardizing the scientific goal of studying preventative treatments. However, in the second instance, if the onset of disease symptoms was used as a humane endpoint, the animal would never develop the disease, the treatment could not be tested, and the scientific goal of the study could never be realized.
Humane endpoints may sometimes seem incompatible with experimental endpoints, because ending an experiment for humane reasons can interfere with achieving the scientific goals of the study. The challenge faced by PIs, veterinarians, and IACUCs is to balance the humane treatment of the animal with the scientific goals of the study. Care should be taken when deciding to terminate an experiment early if this will prevent the study from achieving its scientific goals and thereby potentially wastes the animals. It is equally unacceptable to allow animals to experience pain and/or distress beyond the point required to meet the scientific goals of the study (Wallace, 2000).
Humane endpoints should ideally be based on objective criteria and professional judgment (Toth, 1997, 2000) and should be defined in terms that can be understood and recognized by any staff member coming into contact with the animal. For example, the meaning of the phrase “unable to walk” may be more readily understood that the term “moribund” (Krarup et al., 1999). In situations where diseases may occur spontaneously or unexpectedly (for instance with genetically modified animals), the animal-husbandry staff may be the first to identify a subtle change in behavior or appearance that signals a problem and it is important that they understand and can recognize these changes.
Once humane endpoints are established, they should be defined carefully and thoroughly in the animal-use protocol that is submitted to the IACUC for review. The protocol should also establish an adequate but practical frequency of observations and describe the documentation that will be included in an animal’s health record. The frequency of observations depends on the nature of the experimental manipulation or disease state and the expected rate of change in an animal’s condition.
In some cases, such as genetically modified animals, unpredicted or unintended alterations may occur that adversely affect animal well-being (Stokes, 2000). When developing new types of genetically modified animals, a PI should predict alterations and outcomes based on what is known about the gene of interest, so as to develop humane endpoints (Dennis, 2000). In addition, phenotype screens and measures of general health and well-being may be appropriate to detect unpredicted, adverse alterations in an animal’s physiology. If unexpected outcomes do occur, a change in the frequency of observation or an adjustment of the humane endpoint(s) may be warranted.
Another issue is to identify the individuals who will be empowered to decide that a humane endpoint has been reached and that the animal should be removed from the study and/or euthanized. These individuals should be well trained to recognize what is normal and abnormal for the species, and they should clearly understand what is considered an acceptable or unacceptable condition as specified in the animal-use protocol. A clear designation of authority and responsibility to decide on and carry out euthanasia is essential. Ideally, more than one person should have this authority to accommodate for absences. The IACUC should ensure that a designated contact person is listed in the protocol and that someone will be available for consultation or decisions at all times. The responsible veterinarian must have full authority to carry out humane euthanasia when circumstances warrant, although ideally this should be done after consultation with and with the consensus of the research team.
The Guide (p. 10) states that the method of euthanasia should be considered in the preparation and review of animal-use proposals. The AWRs and the Guide state that the method of euthanasia must be consistent with the current version of the Report of the AVMA Panel on Euthanasia (AVMA, 2001) unless a deviation is justified for scientific or medical reasons. The AWRs stipulate that guidance on appropriate euthanasia techniques be provided to investigators and animal-care staff by the veterinarian (AWR 2.33 (b)(4)). The AWRs also require that records be maintained on dogs and cats that are euthanized (AWR 2.35 (c)(2)). The IACUC must review and approve the method of euthanasia and must determine whether the proposed endpoint of the study is appropriate, inasmuch as the AWRs further require that “animals that are in severe or chronic pain or distress that
cannot be relieved be painlessly euthanized at the end of the procedure or, if appropriate, during the procedure” (AWR 2.31 (d)(1)(v)). The authoring committee notes that requirement does not preclude the development and study of animal models of chronic or persistent pain (AWR 2.31 (d)(1)(iv)(A)); however, animals in severe or intolerable pain should be euthanized. Additionally, animals in studies in which severe pain develops as an unintended consequence should be euthanized or the manipulation causing the unintended pain should be stopped if that would eliminate the pain.
Training staff members to properly perform euthanasia is essential. Training must include instruction both in the specific technique that will be used and in the recognition and confirmation of death (Close et al., 1996). For example, exposure to carbon dioxide can cause deep narcosis that can appear to be, but is not, death. In such cases, animals that superficially appear to be dead may eventually awaken; this arousal can occur after the disposal of carcasses into refrigerators or freezers. The occurrence of death after exposure to carbon dioxide must be confirmed based on careful assessment of the animal for unambiguous signs of death, such as cardiac arrest or fixed, dilated pupils. If an animal is removed from a CO2 chamber before death occurs, the animal either can be returned to the chamber for additional exposure, or, if it is unconscious and nonresponsive, can be humanely euthanized via a physical method (e.g., decapitation or cervical dislocation).
In some species, fear induces animals to become immobile; such immobility must be distinguished from loss of consciousness or death (Close et al., 1996). Some animals release pheromones indicative of fear or distress, which may in turn stress or otherwise disturb other animals (NRC, 1996). Therefore, euthanasia should ideally be performed in an area separate from other animals. However, a recent study suggests that witnessing decapitation may be no more disruptive to Sprague-Dawley rats than other common procedures, such as cage changing, restraint, and injections (Sharp et al., 2003).
Methods of euthanasia that are commonly used in neuroscience research include decapitation, cervical dislocation, carbon dioxide inhalation, and barbiturate overdose. Focused high-intensity microwave irradiation is also used in some cases for measurement of highly labile substances or metabolites (for example, Delaney and Geiger, 1996; Ikarashi et al., 1985; Mayne et al., 1999; Nylander et al., 1997; Theodorsson et al., 1990; Todd et al., 1993). The recommendations of the AVMA Panel on Euthanasia (AVMA, 2001) should be followed unless deviation is justified for scientific or medical reasons (PHS Policy IV(C)(1)(g); APHIS/ AC Policy 3; Guide, p. 65). However, the AVMA Panel consensus concerning the need for anesthetization prior to decapitation is controversial and is based largely on one publication (Mikeska and Klemm, 1975). Other authors dispute the conclusions drawn from that study, concluding instead that hippocampal and cortical responses to decapitation do not reflect consciousness or resemble the response to painful stimuli (Allred and Berntson, 1986; Vanderwolf et al., 1988),
and that instantaneous loss of consciousness, rather than a period of intact pain perception, is likely to occur within a few seconds of decapitation (Bosland, 1995; Derr, 1991; Holson, 1992).
Similarly, there are different views regarding the most humane method for providing euthanasia using carbon dioxide. For example, varying guidance has been provided as to whether it is less distressful to euthanize rodents in a carbon dioxide chamber that has been pre-charged with the gas or not (AVMA, 2001; Close et al., 1996; Hewett et al., 1993; Smith and Harrap, 1997). The most appropriate concentration of carbon dioxide has also been debated, with some authors suggesting that a high concentration promotes a rapid loss of consciousness and death, while others evince that such high concentrations are distressing to the animals (e.g., Danneman et al., 1997). Indeed, recent evidence suggests that carbon dioxide and carbon dioxide-argon mixtures are more aversive to rats and mice than argon alone (Coenen et al., 1995; Leach et al., 2002), although there are a number of reports that carbon dioxide alone provides for a humane death (Hackbarth et al., 2000). Investigators, veterinarians, and IACUCs should be aware of these ongoing debates when they determine the most appropriate method of euthanasia.
The issue of anesthetization or sedation prior to euthanasia is not trivial, because in many circumstances, anesthetizing or sedating an animal before euthanasia, as is recommended for some techniques discussed in the AVMA Panel’s report (AVMA, 2001), has adverse consequences in terms of the validity of the experimental design and interpretation of the resultant data. Because anesthetic and sedative agents exert their effects by altering brain function, use of these agents can alter the concentrations, production, or activity of structures or substances that are being examined to answer an unrelated scientific question (for example, Kasten et al., 1990; Mills et al., 1997; Savaki et al., 1980). In such cases, use of anesthesia or sedation may at worst invalidate the study, rendering the animal experimentation useless, but can also cloud the interpretation of the data, perhaps requiring more animals to be tested. Pilot data to confirm this point may be useful in some cases, but under many circumstances, current knowledge about neurophysiologic mechanisms and metabolic regulation may be sufficient to support the conclusion that use of an anesthetic or sedative would confound interpretation of the data. In such cases, it is the responsibility of the investigator to fully describe in the animal-use protocol the scientific evidence that supports any request to withhold anesthetics or sedatives from animals that are to be decapitated (or cervically dislocated), and it is the responsibility of the IACUC to evaluate this evidence carefully to ensure that it provides a compelling rationale for granting an exception to the recommendations of the Guide and the AVMA Panel.
Animal use in neuroscience and behavioral research usually does not involve the introduction of physical, chemical, or exogenous biologic hazards. However, any animal use involves the potential for an array of suble physical, chemical, and protocol-related hazards and occasional zoonotic disease risks (NRC, 1997). For example, some research programs involve hazards such as the use of the sodium channel blocker tetrodotoxin or 1-methyl-4-phenyl-1,2,3,6 tetrahydropyridine (MPTP) in basal ganglia research.
Laboratory strains of mice and rats are generally free of infectious agents that pose risks to humans. However, other animals used in neuroscience and behavioral research pose zoonotic risks. Examples of specific risks include those posed by wild mammals (hantavirus, rabies, tularemia, and plague), cats (toxoplasmosis and cat-scratch fever), and nonhuman primates (SIV, B virus, shigella, and tuberculosis) (NRC, 1997).
The key to successful handling of experimental hazards is a systematic process for hazard identification during animal-use protocol development and institutional review. Once hazards are identified, risk management should involve the appropriate safety specialists (NRC, 1997).
One class of hazards associated with neuroscience research that merits special attention is the serious and well-recognized zoonotic diseases associated with nonhuman primates. The most problematic are the viral diseases, notably that caused by the macaque monkey’s B virus (also known as Cercopithecine herpesvirus 1). The importance of using awake, behaving rhesus macaques for intensive neurologic study places laboratory personnel at special risk for B virus infection and demands the highest standards of procedural compliance with the use of personal protective equipment, good animal-handling practices, availability of decontaminating equipment, and management of human exposure (Cohen et al., 2002; Holmes et al., 1995; NRC, 2003a). It is essential that laboratories using macaques be well supported by an institutional occupational health and safety program that focuses on the risks of B-virus prevention and control. The basic elements of such a program include procedures and training in dealing with potential exposures, the required use of protective equipment, and access to medical professionals who are knowledgeable about B virus (AAALAC, 2002; CDC-NIH, 1999). Compliance with institutional occupational health and safety requirements should be a prerequisite for IACUC approval of an animal-use protocol and should be evaluated carefully by the IACUC during its semiannual inspections. All macaques, even those from sources thought to be free of B virus and those that repeatedly test serologically negative to B virus, should be presumed to be naturally infected with the virus and handled with appropriate precautions (AAALAC, 2002; NRC, 1997, 2003a).