Animal Welfare Information Center Bulletin, Winter 1999/2000, Vol. 10 No. 3-4

The Triple A Approach to Ensuring Animal Welfare

by
J.A. Davis, DVM
National Institute of Neurological Disorders and Stroke,
National Institutes of Health, Bethesda, MD

Contents

• [Introduction]

• Assessment of discomfort in an animal model of experimental autoimmune encephalomyelitis

• Acknowledgment

• References

• Tables

• Related article: Strategies for Providing Interventional Nursing Care

Russell and Burch’s (1959) Three Rs (reduce, refine, replace) provide the framework for reducing animal pain and distress in biomedical research. Replacement has, perhaps, gained the most attention because activist groups targeted toxicity testing in animals. However, toxicity testing represents only a small percentage of research involving animals. Refinement, on the other hand, has the greatest potential to reduce animal distress and/or pain due to experimental procedures and frequently reduces the number of animals used.

Good experimental design defines the variables of interest. All remaining parameters remain the same in order to isolate and assess the specific variable(s) of interest. Fundamentally, the experimental design is at risk of compromise if confounding variables are introduced through poor animal well-being. These unintended consequences (either physiological, psychological, or both) affect the scientific data collected and thus should be avoided as much as possible. Avoiding such unintended effects on animal well-being involves much more than selection of the appropriate anesthetic or analgesic; good experimentation demands employing strategies that minimize factors that would direct physiological (and psychological) responses from the norm. Attention given to refinement from this point of view will also result in other benefits such as strengthening the scientific data (less inter-animal variability) and should also lead to reduction in numbers of animals used because animal morbidity and/or moribundity usually lead to greater numbers of animals needed to reach the scientific objective.

In this article, I use the term “refinement” in the sense of finding methods to relieve potentially painful or distressful procedures. I will focus on refinement of protocols in which the animals may not show overt signs of pain or distress as typically defined but may be considered abnormal because of the experimental disease that is part of the experiment. For example diabetes mellitus, induction of tumors, anemia, etc. are diseases that would be treated clinically but under experimental protocols, it may only be possible to alleviate symptoms. Because it is  difficult to determine if an animal feels pain or distress, we (the veterinary and scientific staff) try to assess various procedures for their potential to disrupt the animal’s physiological responses. The problem with assessment is lack of specific indicators of pain and the subjective nature of the assessment system. The clinical signs used in our facility to assess the condition of the animal were assumed to indicate early clinical disease; therefore, intervention strategies were expected to stabilize or slow progression of disease. We have no reliable indicators of pain or distress and are left solely with clinical assessment. It is our contention that early intervention and treatment of clinical signs does refine the study from both the viewpoint of the animal and the investigator. Daily observation, assessment, and intervention allow animals to remain stabilized longer, gain weight, regain bladder function, etc. For the investigator, such treatments allow experimental groups to  remain intact until the endpoint of the study without compromising study objectives. Before our staff began such a system, a large percentage of animals had to be euthanized because of  complications from the disease before reaching the study endpoint, which frequently necessitated replacement animals.

At NINDS, in developing assessment strategies, we believe it is important that the scientist and veterinarian form a partnership because, to be successful, it must be a dynamic process. As each partner gains experience with the animals’ responses to the experimental manipulations, we are better able to refine our intervention strategies, predict earlier intervention points, and apply our experiences to other similar protocols. We stress that the process is dynamic because we are rarely able to precisely predict how an animal model will behave. As the adage about hindsight suggests, only through experience are we able to clearly see errors made in our predictions of animal response in early stages of experimental design. However, rather than wait until a study is ended and reviewed, we believe making adjustments on intervention strategies and treatments can lead to better outcomes for both the animal and the scientist. We do not attempt to detect distress or pain—rather we rely on our knowledge of the animal’s normal behavior and decide that any observations that deviate from normal behavior indicate the need for some type of intervention. In order for assessment to work, it is essential that the evaluators know normal physical appearance and pattern of behavior of the species in a study. In fact, it is often an investigator’s lack of knowledge of the animals’ normal behavior that can lead to disagreement in early phases of constructing assessment charts. However, in our experience, directly involving the investigator in animal observations quickly builds confidence  in the partnership to correctly interpret an animal’s overall condition.

Admittedly, judgments are subjective for the most part; however, as one gains experience with the animal model, acuity increases. A comatose or moribund state in itself may not be painful or distressful to the animal but earlier signs may have been missed. We collect both objective and subjective data to assess the animal’s well-being. Examples of objective data are temperature, body weight, blood and urine glucose, blood pH, respiratory pattern etc; examples of subjective data are appearance, posture, and response to stimuli.

We suggest that there are two phases to the study: a planning phase in which the approach is defined and an execution phase in which the chosen approach is continually assessed and adjusted (our 3 A’s). The sum of these activities results in assurance that we are eliminating or mitigating as much animal distress and/or pain as possible.

 [Contents | References]

Assessment of discomfort in an animal model of experimental autoimmune encephalomyelitis

In one noteworthy example, we had the opportunity to assess the degree of possible discomfort in a mouse model of experimental autoimmune encephalomyelitis (EAE) , which creates an acute phase followed by relapses with intervening periods of remission due to lesions in the brain and spinal cord (demyelination and inflammation). Following the acute phase, which is fairly synchronous among mice injected on the same date, relapses occur on different days (asynchronous), often resulting in an underestimate of clinical severity.

Depending on a study’s objectives, investigators may assess differences between day of onset, mean peak of disease onset, number of relapses, and mean clinical scores. During initial review, investigators did not provide an assessment intervention chart for the clinical course of disease. Rather, they provided a clinical grading scale that had been previously published to objectively measure severity of disease (table 1). The veterinary staff wanted a different chart that we could use as a guide to provide nursing care interventions or determine if euthanasia endpoint criteria existed. Concerned that the categories were too general to prove useful from a veterinary viewpoint, we met with the investigators to develop a mutually acceptable grading scheme that would meet study objectives, establish guidelines for intervention, and not interfere with study goals. Initially, this idea met with resistance. Concerns of the investigators were that we would be overzealous in our assessment and euthanize animals without discussion or we would choose inappropriate (for the study) drugs to treat animals. After discussion, an agreement was reached that resulted in table 2. The veterinary staff agreed not to euthanize an animal unless all attempts to reach the investigator had been made and, in the opinion of the veterinarian, the animal was approaching death. We also agreed to save tissues from animals we euthanized. Last but not least, we agreed to a list of drugs that could be used that would not interfere with the study and all intervention therapies were approved in advance, as were criteria for euthanasia. The investigators, on the other hand, agreed to write the date of injection of sensitized lymphocytes on the cage card plus the animal’s weight (considered baseline weight). They also notified us as to when they expected clinical signs to begin. It was anticipated that the information obtained would provide guides for the feasibility of assessing discomfort in EAE animal models. Discomfort was assessed quantitatively by veterinary technicians and three veterinarians who were familiar with the experimental animals. Female mice (SJL/J) were used as this strain is recognized as being susceptible to induction of EAE. The facility is maintained as a specific-pathogen free facility and cages are maintained in environmentally controlled rooms, at 72^o ± 2^o F and 40-50 percent humidity, on a 12:12 hour light:dark cycle, with lights on at 6 AM (0600). All mice were fed a standard laboratory diet (NIH-07 Rat and Mouse Chow, Ziegler Bros., Gardners, Pennsylvania) and provided water ad libitum. Mice were kept in polycarbonate cages with a layer of bedding (Tek-Fresh, Harlan TEKLAD, Madison, Wisconsin) and filter tops in a ventilated rack (Maxi-Miser Caging System, Thoren, Hazelton, Pennsylvania). At the time of injection with myelin basic protein (MBP), the mice were weighed.

As the veterinary technical staff began to use the assessment chart, we began to recognize subtle impairment in motor skills. This was a direct result of pulling each cage of mice to closely observe them as we approached the time that the investigator had estimated clinical signs might begin. As we developed our observational skills, we began categorizing mice in EAE grade 1 earlier and earlier, which resulted in more intense monitoring. As we gained both experience with the animal model and investigator confidence, we began modifying the assessment chart to reflect more intense nursing during acute and relapse phases of the disease. The assessment charts became increasingly more complex (tables 3 and 4). The results of this were:

• Animals lived longer allowing investigators to reach study endpoints.
• There were anecdotal stories that the investigators did not believe their animal data because animals lived longer and maintained fairly stable weights (except during acute phases of the disease).
• Our experience with the mouse model helped us formulate similar assessment charts for EAE studies with marmosets and pigs.
• The investigators now ask for our observational data (weight) to correlate with their measurement of disease.

Through persistence, a partnership forged collaborative recognition and reduced the number of animals used for each experiment.

Our experience, over the past several years, in refining these interventional charts is worth noting. First of all, we learned that before we began to pull individual cages, it was best to look at the cages from a distance. By this, I mean we learned to scan cages on a rack, looking for differences in animal activity from cage to cage. We were able to quickly identify new cages of animals showing clinical signs by decreased activity or inability to see the animals (partially buried in bedding or remaining at the back of the cage). We also learned to begin handling the animals as soon as they arrived to acclimate them to the increased handling and observation that was sure to follow once an experimental group was formed. This point cannot be overemphasized—acclimating animals to handling/observation minimizes measurable differences in biological responses that may adversely affect the experiment (stress). This also allows the staff to experiment with food enrichment which, during later illness, may be invaluable in providing needed caloric intake. Second, we pulled cages and observed animal activity and appearance. Subtle changes in overall appearance of a group of animals may be the first indication of abnormality. We concentrated primarily on motor function, such as how the animal rested: hunched up (abdominal pain) or legs in abnormal positions (straight out behind them = paresis; wide front leg stance = chest pain or difficulty breathing) or on animals that seemed hesitant to move (gait disturbance), animals isolated from the group, etc. These mice were individually assessed.

Figure 1. A) Plantar reaction from a normal mouse [left]. B) This mouse [right] will appear normal in the cage but with you pick it up, he exhibits "clasping" rather than the normal plantar reaction.

If you pick a mouse up by the base of its tail, a normal mouse will display a typical plantar reaction (figure 1a) whereas mice with early signs of neurological deficit exhibit a clasping behavior (figure 1b) that can be unilateral (less severe) or bilateral. Other signs of onset of clinical disease were bradykinesia and/or hypokinesia, that were usually unilateral and disappeared, initially, when the mice were stimulated to walk. Therefore, it was important to be particularly observant of the animals on initial opening of a cage or early signs of neurological deficit might be missed. When mice exhibited any mild deficit, they were weighed and classified as EAE grade 1. From that point forward, mice in this cage (and injection group) were observed more stringently and mice with neurological deficits were closely examined each day for progression of disease. Another assessment made was noting how well mice groomed themselves. As EAE progressed, we began to see urine scald and or evidence of fecal material around the anal orifice. This observation led to the need to clean these mice at least twice daily. Other physical assessments made were palpation of the bladder—frequently the bladder became atonic, requiring urine expression 2-3 times daily—and assessment of dehydration by noting skin turgor, coat condition, recessed eyes and occasionally measuring packed cell volume and specific gravity of urine. As Beynen et al. (1987) reported, we found that this approach to health assessment of EAE animals led to earlier intervention points, more intense nursing care than originally envisioned, and modification of the EAE clinical charts. We also became more appreciative of the variability of response between animals within a treatment group and within a cage. It was the rule, not the exception, that at least one or more animals rarely progressed to EAE 3 during the acute stage of the disease — most were EAE 2; however, an occasional animal did qualify for EAE 4 designation, which required intense nursing. We agree with Flecknell (1994) that we cannot blanket treat animals. Significant refinement of procedures will occur only if individual animal assessment is conducted as standard procedure. We also weighed the mice with increasing frequency as clinical symptoms indicated reduced food intake. Other signs of reduced food intake, such as decreased urine/fecal production are more difficult to assess in a cage of animals, unless all animals are in the same clinical condition. However, we did use soiled bedding as an indicator during our initial scan of cages from a distance. When mice were designated EAE2/3 or EAE 3, we began assessing their cardiovascular system by noting cold extremities and pale ears. Heat supplementation was provided.

The above example is only one way in which we try to refine animal experiments; our goal is to refine as many procedures as possible. This may mean convincing investigators to switch from injectable anesthetics to inhalant anesthetics, which frequently have fewer side-effects on animal physiological responses. Similarly, attention to intraoperative monitoring of anesthesia can significantly refine a study by minimizing detrimental effects and reducing risk of over-anesthetizing animals which leads to their death. We also consider the method of anesthetic or analgesic delivery, realizing that their use is not without its own inherent stress to the animals—and thus, as stated earlier, there is an emphasis on conditioning animals to handling prior to experimentation. Other areas that we are working on are refinement of tumor studies, use of complete Freund’s adjuvant, and ascites production of monoclonal antibodies. We recognize that there are many ways to achieve refinement if we learn to look beyond our own assumptions and conditioning as to what is “normal” in the way we interact with animals.

 [Contents | References]

Acknowledgment

I would like to thank Dr. Victoria Hampshire for her creativity in finding solutions to provide excellent nursing care to these mice.  She experimented with several gelatin recipes before finding one that the mice ate and provided the caloric content needed to maintain or increase their weight. She also created the “mouse pocket.”

 [Contents | References]

References

Beynen, A.C., V. Baumans, A.P.M.G. Bertens, R. Havenaar, et al. (1987). Assessment of discomfort in gallstone- bearing mice: a practical example of the problems encountered in an attempt to recognize discomfort in laboratory animals. Laboratory Animals 2l: 35-42.

Flecknell, P.A. (1994). Refinement of animal use—assessment and alleviation of pain and distress. Laboratory Animals 28: 222-231.

Russell,W.M.S. and R.L. Burch (1959). The Principles of Humane Experimental Technique. Universities Federation for Animal Welfare: Hertfordshire, England, 238 p.

 [Contents | References]

Tables

Go to:
Contents, Animal Welfare Information Center Bulletin
Top of Document

The Animal Welfare Information Center
U.S. Department of Agriculture
Agricultural Research Service
National Agricultural Library
10301 Baltimore Ave.
Beltsville, MD 20705-2351

Phone: (301) 504-6212
FAX: (301) 504-7125
Contact us: http://www.nal.usda.gov/awic/contact.php

Policies and Links