CCAC guidelines on: choosing an appropriate endpoint in experiments using animals for research, teaching and testing

1. INTRODUCTION
   
   The CCAC policy statement Ethics of Animal Investigation states:
   
   "Animals must not be subjected to unnecessary pain or distress. The experimental design must offer them every practicable safeguard, whether in research, in teaching, or in testing procedures; ..."

   (Ethics of Animal Investigation, CCAC, 1989)

   
   The Ethics of Animal Investigation, also requires investigators to follow the "Three Rs" of Russell & Burch (1959): Replacement (of animals with other, non-sentient material or with animals of lower sentience); Reduction (of numbers of animals used); and Refinement (of technique, "to reduce to an absolute minimum the amount of distress imposed on those animals that are still used.").
   
   The investigator's ethical responsibilities are clearly stated in this policy statement; however, some important questions need to be answered. How can an endpoint be chosen that satisfies these principles? How can experiments be "refined" through establishing earlier, more humane endpoints to invasive animal experiments, particularly those that may have used death as an endpoint in the past? Where should the line be drawn?
   
   For the purposes of these guidelines, the term "Endpoint" is defined as the point at which an experimental animal's pain and/or distress is terminated, minimized or reduced, by taking actions such as killing the animal humanely, terminating a painful procedure, or giving treatment to relieve pain and/or distress.
   
   There are several types of studies where the death of the animal may in the past have been the endpoint as part of the experimental design, but where this requirement is now questioned. These areas include regulatory toxicology, diagnostic toxicology, acute toxicity studies in research, infectious disease studies, microorganism virulence challenge studies, vaccine efficacy trials, cancer research, and cancer treatment evaluation.
   
   In some research investigations, pain and/or distress is an unwelcome part of the disease or condition being studied (e.g., some models of human diseases such as arthritis or cancer). Also, in some experimental animal use, any pain and/or distress is a side effect to the animal use (e.g., monoclonal antibody production, Freund's Adjuvant use in antibody production). In these latter cases, endpoints are relatively easy to define (e.g., limiting the volume and number of times ascites fluid is collected from a mouse), and a policy statement pertaining to some of these procedures already exists (CCAC, 1991).
   
   
2. GENERAL GUIDELINE
   
   In experiments involving animals, any actual or potential pain, distress, or discomfort should be minimized or alleviated by choosing the earliest endpoint that is compatible with the scientific objectives of the research. Selection of this endpoint by the investigator should involve consultation with the laboratory animal veterinarian and the animal care committee.



3. RECOMMENDED PROCEDURES FOR SELECTING AN APPROPRIATE ENDPOINT
   
   The animal in a moribund state may be past suffering (and actually comatose). A moribund animal is one that is close to death and may be comatose or unresponsive to stimuli, exhibit dyspnea or other severe breathing problems, hypothermia, prostration, etc. However, before the animal gets to the point of being moribund, detailed observations of the animal can help to set an earlier endpoint and thereby reduce the actual cost to the animal, in terms of pain and/or distress.
   
   There are several considerations in defining an appropriate endpoint in a given experiment. These all depend on an objective determination of any deviations from an animal's "normal" state, followed by a correlation of these changes with degrees of discomfort, pain and/or distress. Some of these considerations are:



• making the appropriate observations of the animals (of their behavior, physiology, etc.);


• assigning objective values to the observations of animal behavior and physiology;


• determining which observations are the most significant indicators of pain and/or distress in the specific circumstances of the research;


• determining which observations are the most significant predictors of further deterioration in the animal's condition, and then identifying the earliest point at which those signs appear;


• meeting the scientific demands for an objectively measured and significant endpoint;


• clearly defining the information/data being sought in the experiments.


1. Making the Appropriate Observations of the Animals
   
   Morton & Griffiths (1985) laid the groundwork for developing a set of observations for assessing pain, distress and discomfort in laboratory animals, based on evaluating five aspects of an animal's condition:



• changes in body weight (and related changes in food and water intake);


• external physical appearance;


• measurable clinical signs (e.g., changes in heart rate, in respiratory rate, and in their nature);


• changes in unprovoked behavior; and,


• behavioral responses to external stimuli.

In each of these categories, a rating system of 0 (normal or mild) to 3 (severe changes from normal) was proposed. The cumulative rating obtained by adding the score for each category indicates increasing deviation from normal in the animal, which can be interpreted as an indication of increasing pain and/or distress. A total score can be identified, at which point the animal's pain and/or distress will be terminated or alleviated. The observations listed by Morton & Griffiths (1985) are only some of the observations that may be required to determine an endpoint in specific research models.

In addition to behavioral and physiological measurements, a number of hormonal indicators have been used to measure stress/distress in animals (catecholamines, corticosteroids, prolactin, tumor necrosis factor, interleukins) (NRC, 1992). There is little consensus on which of the many possible hormonal changes truly measure distress or pain. Nevertheless, if blood sampling is part of the research protocol the investigator should consider analysis of the blood collected for some of the stress indicators. This information would be a valuable addition to the data, and could provide a useful correlation to any noted behavioral changes.

It is important to be aware of the characteristic behavior of the species under observation. Animals such as non-human primates, rodents, rabbits and some livestock may not show many behavioral changes even when in severe pain. In addition, strain variations must be considered. It is, therefore, imperative that the investigator/observer understand these characteristics prior to setting endpoints.

In practical terms, the animal should first be observed carefully without disturbance. Its appearance and posture should be observed, and a determination made about whether its behavior is normal or abnormal. The animal's reaction to an external stimulus could also be checked (e.g., noise, change in light level) before directly approaching the cage to handle the animal. Depending on the species, the animal should be handled for a clinical examination. During this time, clinical signs are noted and measurements made (including weighing the animal). Any lesion or abnormality is also evaluated at this time (e.g., size of tumor). During the clinical examination, "provoked behavior" can also be assessed.

The use of observational "checklists" for scoring the animal's condition in a study provides an objective basis on which decisions about endpoints can be made. The advantages of checklists are that specific observations are not overlooked or taken for granted. The other real advantage is that such checklists help improve observational skills, particularly with the smaller laboratory animals where some of the conventional clinical observations made on larger animals are not readily determined (e.g., temperature, heart rate, respiratory rate). However, checklists do not cover all abnormalities or observations and thus are only a useful way to record certain findings. They cannot replace a thorough examination of the animal.

Morton & Griffiths (1985), and Sanford, et al. (1986), focussed attention on the need for more objective assessments of the pain and/or distress that may occur in an animal in the course of biomedical research. Efforts at refining the scoring of clinical signs have continued since 1985 (Morton, 1990; Morton & Townsend, 1995; Workman, et al., 1998, for example). More detailed observational checklists have been proposed for some specific scientific procedures, including: endotoxin administration in mice (Townsend & Morton, 1994); monoclonal antibody production in mice (Morton, 1997); cancer research (Workman, et al., 1998).

In addition to general signs of pain and/or distress, there are specific signs and symptoms related to the condition being studied. For most animal models of disease, information on organ system(s) affected, specific symptoms, progression of symptoms, time course of the disease condition, and expected lesions, is available from the comparative medicine and general veterinary literature (e.g., Armed Forces Institute of Pathology [AFIP] Fascicles). Such specific signs and symptoms must also be used in the overall evaluation of the animal's condition.

The result of previously published information or pilot studies, as well as information about the pharmacology or chemistry of the compound being tested should be used to predict any potential adverse effects on the animals. Implicit in this is a requirement that a complete literature search has been conducted.


2. Scoring of Significant Physiological Observations to Select and Refine Endpoints
   
   Information on the general signs of pain and/or distress for the various animal species commonly used in biomedical research are readily available (CCAC, 1993; Sanford, et al., 1986; Wallace, et al., 1990). Of these signs, significant weight loss may be one of the more important signs of deterioration in the animal's condition (reflecting a change in food and water consumption). Weight loss in these circumstances must always be compared to the appropriate control animal. Body condition scoring charts, which are available for domestic livestock, and dogs and cats, may be useful for evaluating chronic weight loss in experimental animals in specific studies.
   
   Hypothermia can also be an important indicator of a deteriorating condition in the animal, when it occurs in specific disease or toxic states. Wong, et al. (1997) found that a decrease in body temperature (below 32^oC) of mice infected with an influenza virus was predictive of mortality. Soothill, et al. (1992) found that in mice infected with bacteria, hypothermia of 34^oC was predictive of mortality. Thus, in specific experimental cases, the point at which the body temperature of an experimental animal drops to a specified temperature could be set as the endpoint at which euthanasia is recommended.
   
   The report of a committee of the British Laboratory Animal Science Association (Wallace, et al., 1990) includes an assessment of the severity of some procedures commonly performed on animals in the course of biomedical research. A number of other publications are also available to help identify the signs and symptoms of experimental animal pain and/or distress (Barclay, et al., 1988; Baumans, et al., 1994; British Veterinary Association, 1985; Butler, et al., 1985; De Castro Costa, et al., 1981; Flecknell, 1994; Keefe, et al., 1991; Wolfensohn & Lloyd, 1994; Soma, 1987).



3. Identifying Significant Behavioral Indicators of Pain and/or Distress
   
   For any given animal model, there are many possible observations and measurements that can be made. Determining which are the most important or significant indicators of the condition of the animal, or perhaps more importantly from the investigator's perspective, which are the most important indicators of an irreversible deterioration in the condition of the animal, is not an easy task. Studies by Butler, et al. (1985) and De Castro Costa, et al. (1981), dealing with the adjuvant-induced arthritis model in the rat, provide insight into the difficulties in finding/choosing the correct observations. Although changes in the frequency of several behavior patterns were found (decreased rearing, running, eating, drinking and climbing; increased resting, freezing, scratching), the conclusion was that of all these changes in behavior, increased scratching was the most significant behavioral change that tied in with developing arthritis, indicating chronic pain.
   
   Such behavioral evaluations (Butler, et al., 1985; De Castro Costa, et al., 1981) are research projects in themselves, involving many hours of technical time with expensive monitoring and analytical equipment. It may be unrealistic to demand a similar degree of preliminary evaluation each time an animal-based research program is initiated where the potential for pain and/or distress is high (mice in a liver cancer research program, for example). Nevertheless, conducting a pilot study to establish the observational criteria to be used to set endpoints may be a very useful exercise, particularly at the onset of a research program.


4. USING PRELIMINARY OR PILOT STUDIES TO DETERMINE THE APPROPRIATE ENDPOINT
   
   The use of preliminary or pilot experiments can be very useful in determining endpoints (Olfert, 1995; Browder, 1995, Everitt & Griffin, 1995), particularly when the effects of the treatment on the animals are unknown. A pilot study, using a small number of animals, may help determine the morbidity, time course of effects, and frequency of observations required to set an earlier endpoint. A pilot study can also provide an indication of the variance of responses between treatment groups, which can then be used to estimate group sizes more accurately for the main study. Conducting a pilot experiment also provides the opportunity for all persons to become experienced with the expected signs and symptoms.



5. DETERMINING THE REQUIRED FREQUENCY OF ANIMAL OBSERVATIONS
   
   Guideline: Based on previous knowledge, during critical periods of the experiment and at the onset of adverse reactions, a minimum of two or three observations should be made daily. The frequency of the observations should increase depending on the potential for increasing pain and/or distress.
   
   The CCAC Guide to the Care and Use of Experimental Animals states that normal, healthy experimental animals should be observed at least once a day (CCAC, 1993). However, once an animal is in a potentially critical period with respect to impairment, more frequent observations must be made.
   
   The frequency with which affected animals should be observed must be determined for each study. The required frequency of, and interval between, observations will depend on the expected or known time course of the condition. For example, in some experimental infections/toxicity cases, much more frequent (e.g., hourly) observations may be necessary to identify the point at which the selected "endpoint" has been reached and the animal's pain and/or distress must be terminated (Morton & Townsend, 1995; Townsend & Morton, 1994). Scheduling the study so the critical period for the animals occurs during normal working hours (when the lights are on in the animal room) may aid in ensuring that appropriate observations are made.
   
   The appropriate monitoring schedule should be established by the investigator in consultation with the veterinarian, and approved by the ACC in its consideration of the protocol.



6. DEFINING RESPONSIBILITY FOR ANIMAL OBSERVATION
   
   Guideline: With respect to setting and determining endpoints, the responsibility(ies) of each individual should be clearly defined, and a clear chain of reporting established. The ultimate authority for euthanasia must rest with the veterinarian, supported by the animal care committee.
   
   It is essential that an appropriately trained and experienced person with the authority to euthanize, or order the euthanasia of, animals be constantly available during the study. Any observations of unusual behavior or signs of pain and/or distress should be reported immediately to that authority. This is important also for dealing with unanticipated adverse effects on the animals in an invasive study. The authority to euthanize animals that have reached the endpoint, or are experiencing severe unanticipated adverse effects should be clearly defined before the study begins. The ultimate authority must remain with the institutional veterinarian (CALAM, 1990).



7. TRAINING OF PERSONNEL IN CLINICAL ANIMAL OBSERVATIONS
   
   Guideline: All persons responsible for making observations of the animals, from which an endpoint will be determined, should be competent in evaluating the normal physiology, behavior and body condition of the animals under observation, and the anticipated specific changes from normal.
   
   It is the responsibility of the principal investigator or study director (and ultimately the ACC) to ensure that all persons involved have the training appropriate to their responsibilities for animal observation. The training should be documented.



8. THE ROLE OF THE INSTITUTION'S ANIMAL CARE COMMITTEE IN SETTING ENDPOINTS
   
   The role of the ACC is vital in establishing the structure to ensure that the earliest endpoints consistent with producing reliable data are considered, identified, and used. This is a joint responsibility with the investigator and the veterinary staff. ACCs should obtain information on the following questions, to ensure that an appropriate endpoint will be in place:



• what are the scientific justifications for using the proposed endpoint?


• what is the expected time course for the animals, from initial treatment to first signs of pain/distress, to the death of the animal, based on previous information with the specific model under study?


• when are the effects to the animal expected to be the most severe?


• if the course of the disease and expected signs of the adverse effects are unknown, could an initial (pilot) study, under close observation by the investigator and/or laboratory animal veterinary staff, answer these questions?


• has a checklist of observations, on which the endpoint will be based, been established?


• who will monitor the animals (identify all responsible) and keep records?


• has a clear chain for reporting observations been established?
  
• what will be the frequency of animal observations: a) during the course of the study; and b) during critical times for the animals?


• do the investigators, animal care and technical staff have the training and expertise to monitor the animals adequately?


• what provisions have been made to deal with any animals that show unexpectedly severe signs and symptoms?


• for toxicological studies, have existing toxicological data been evaluated?