the Goal Justify the Methods? Harm and Benefit in Neuroscience Research Using Animals

Replacement is the first of the three Rs, for several reasons:

Replacement enjoys a particular standing among the three Rs. It was the first of the Rs to be introduced by Russell and Burch (1959), reflecting the intended order in which the Rs were to be considered. Questions about Reduction and Refinement are only relevant if Replacement has first been considered and excluded. The goal of Replacement also has received widespread support, in part because it is the only goal that is fully compatible with the animal rights perspective that animal use solely for human benefit should not be permitted (Olsson et al. 2011).

The main point of this principle of Replacement is that the use of animals should be replaced by nonanimal methods whenever this is possible without compromising the research objective. Replacement methods can be divided into four main types: in vitro (e.g., cell lines), ex vivo (e.g., tissue cultures), in silico methods (e.g., bioinformatics), and research with human volunteers. The idea that studies in human volunteers would be an ethical alternative to the use of animals in research may seem provocative, and it is of course a sine qua noncondition that such a study meets the ethical standards for research with human subjects. That said, in the neurosciences there seems actually to be real potential for this approach where replacement not only spares animals but also increases the relevance of the research itself. Focusing specifically on pain research, a workshop which brought together neuroscientists with proponents of nonanimal research methods came up with a number of suggestions for situations in which studies with human volunteers could replace animal studies. They are all based on the use of low-risk minimally invasive techniques (e.g., functional imaging and microdialysis) in humans, and draw on the fact that it is much easier to evaluate subjective experience—a key aspect of pain—in verbal humans (Langley et al. 2008).

Another example of a replacement strategy in neuroscience research was recently described by Barnett and collaborators (Sorensen et al. 2008; Boomkamp et al. 2012). These authors proposed an in vitro method for research on spinal cord injury, a disorder that has depended mainly on animal research. Spinal cord injury is a complex injury, caused by traumatic accidents. Traumatic injury disrupts spinal white matter tracts, resulting in loss of sensory and motor function. This loss of function is generally permanent because the central nervous system has a restricted regenerative capacity (Fawcett and Asher 1999; Rudge and Silver 1990). After the initial injury, which results from direct mechanical disruption of spinal cord integrity, glial scars are formed, which inhibit central nervous system repair by creating both physical and biochemical barriers to axonal growth (Boomkamp et al. 2012).

An example of an animal model of spinal cord lesion is a wire knife lesion, generated by inserting the knife into the dorsal column and pulling up a piece of tissue. The method results in a cavity and glial scarring that mimics human spinal cord injury. Disadvantages to rat models of spinal cord injury include the need for large numbers of animals, the severity of the procedure for the animals, the long time frame for results, and the high expensive of the experiments (International Animal Research Regulations 2012).

In the nonanimal model proposed by Barnett and collaborators, embryonic spinal cord cells from rats are layered on top of an astrocyte monolayer derived from embryonic tissue. Growth in culture over time leads to complex axonal/glial interactions resulting in myelinated neurons. This system allows for the study of contact between astrocytes and how they communicate with the axons, which is necessary for understanding the problems in spinal cord injury. The researchers also have induced lesions in the cell culture by cutting with a scalpel to studying axon density and myelination adjacent to the lesion and cell growth into the damaged area (International Animal Research Regulations, Impact on Neuroscience Research, Workshop Summary, Institute of Medicine (US); National Research Council (US) 2012).

Overall, the greater the role of nonanimal replacement in research, the fewer animals will be needed in total for research purposes. In this way, replacement is also directly related to the second R, Reduction.

The aim with the principle of Reduction is to use the smallest possible number of animals to obtain valid information. Its main ethical purpose is to reduce collective animal harm, understood as the number of animals on which harm is inflicted. One important measure is to use correct and careful statistics, namely by carrying out appropriate power analysis prior to study commencement. Sample sizes can also be decreased by controlling variance associated with different environmental and genetic conditions, as for example, by using uniform housing conditions and inbred animals.

Reduction is probably the most controversial of the three Rs. There is a great political value in bringing down numbers of animals used in experimental procedures as a whole, as the number of animals reported in annual statistics is a very visible and easily understood aspect of research animal ethics . This also holds for replacement—performing fewer experiments is also immediately recognizable in the statistics. However, the problem with reduction is that, as detailed analyses have repeatedly shown, in actual research the number of animals used in an individual experiment is often too small for results to be reliable. This of course has important implications for the validity of the research results. Within a larger review of methods in neuroscience , Button and collaborators (2013) examined the statistical power of animal experiments investigating sex differences in water maze and radial maze performance. The effect (i.e., how large a difference is between male and female animals) was calculated through a meta-analysis, and the authors then established how many animals a single study would need to detect effects of this magnitude with different levels of statistical power. To achieve 80 % power (a common standard), 134 animals would be needed for a water maze experiment and 68 for a radial maze, whereas the average sample sizes were 22 and 24 animals, respectively. The authors commented on the ethical consequences of underpowered studies:

There is ongoing debate regarding the appropriate balance to strike between using as few animals as possible in experiments and the need to obtain robust, reliable findings. We argue that it is important to appreciate the waste associated with an underpowered study—even a study that achieves only 80 % power still presents a 20 % possibility that the animals have been sacrificed without the study detecting the underlying true effect. If the average power in neuroscience animal model studies is between 20–30 %, as we observed in our analysis above, the ethical implications are clear.

Low power therefore has an ethical dimension—unreliable research is inefficient and wasteful. This applies to both human and animal research. The principles of the ‘three Rs’ in animal research (reduce, refine and replace) require appropriate experimental design and statistics—both too many and too few animals present an issue as they reduce the value of research outputs.

Based on this, it does not seem appropriate to apply reduction through uncritically decreasing sample sizes in individual experiments. Additional approaches in experimental design are needed if the aim is to bring down animal numbers. This could include the use of imaging techniques allowing the study of disease progress in the same animals rather than in separate groups for separate time points, or greater use of nonanimal approaches before moving to an animal model .

Whereas the Replacement and Reduction principles reduce harm by avoiding the use of animals, the Refinement principle addresses the welfare of individual animals which are actually used in experiments. This principle states that all experimental procedures shall be adjusted to minimize any pain or discomfort they may cause to the animals. Experiments can be refined in several ways, from the use of anesthesia and analgesia, to housing adaptations and the establishment of human endpoints. Appropriate measures need to be defined for each individual study, taking into account the nature of the harms which need to be mitigated. The scheme for welfare assessment recently proposed by a European working group allows refinement measures to be integrated into the assessment. Table 2 displays examples of refinement measures that can be applied in neuroscience studies whose harms were presented in Sect. 2.2 (and are now summarized in the scheme).

The principle of the three Rs is already present in much legislation. For example, the new European Directive states that “To ensure that the way in which animals are bred, cared for and used in procedures within the Union is in line with that of the other international and national standards applicable outside the Union, the principles of replacement, reduction and refinement should be considered systematically when implementing this Directive.” Although the three Rs principle was not explicitly referred to in previous European legislation, researchers were asked to use animals only when necessary, to use as few animals as possible and to use procedures having as little impact as possible.

Unfortunately, systematic reviews of the implementation of refinement measures in biomedical research indicate that the present situation is far from ideal. For example, between 2000 and 2002 pain relief was administered in only around 20 % of studies subjecting rodents to potentially painful procedures (Richardson and Flecknell 2005). In 2009, humane endpoints were only reported in about 20 % of studies of mice models of the neurodegenerative disorder Huntington’s Disease, with no significant increase in the reporting of this refinement measure during the preceding 10-year period (Franco and Olsson 2012). There is thus considerable potential for improvement in the application of refinement.

In this final section regarding harm, we will address an idea that is recurrent in the discussion of ethical evaluation and regulation of research: that research will be more or less harmful depending on the animal species chosen (see also Chapter “Would the Elimination of the Capacity to Suffer Solve Ethical Dilemmas in Experimental Animal Research?​” of this book). There are sometimes obvious physical justifications having to do with the size of the animal in relation to the minimum amount of tissue needed for analysis or the minimum size of lesion determined by human dexterity and instruments used—in such cases the smaller the animal the larger the proportional impact will be.

But there is also the widespread idea that animals of different species vary in their capacity for subjective experience. This idea is put forward in the European Directive which requires that if several methods are available one shall choose those that “involve animals with the lowest capacity to experience pain, suffering, distress or lasting harm” (Directive 2010/63/EU, Article 13). This seems to indicate that animals can be different on their capacity to suffer—but no guidelines are given for how to assess this capacity. Smith and Boyd (1991) proposed a systematic method of assessment consisting of a checklist of neuroanatomical/physiological and behavioral criteria to determine whether a nonhuman animal has the capacity for pain, stress, and anxiety. On the neuroanatomical side the criteria include (1) the possession of receptors sensitive to noxious stimuli (nociceptors), (2) the possession of higher brain centers (especially a structure analogous to the human cerebral cortex), (3) the possession of nociceptors connected to these higher brain structures, and (4) the possession of opioid-type receptors. On the behavioral side the criteria include (5) responses to painful stimuli modified by analgesics, (6) avoidance or escape responses to painful stimuli, (7) responses to noxious stimuli that persist, and (8) the capacity to associate neutral with noxious stimuli. However, looking at how taxonomically distinct animals used in research fare in this assessment it becomes clear that (1) if we complement the information available in the original 1991 analysis with contemporary knowledge about fish, at least all vertebrate animals meet the criteria for pain and (2) information about nonvertebrate sentience is too limited to allow species to be identified as less sentient with reasonably reliability. That is, it is highly unclear what animal researchers are to choose to ensure “lowest capacity to experience”. Colin Allen (2004) proposed to use learning abilities as indicators of capacity to suffer. This would include operant learning—which appears to require a brain (Grau 2002), and certain kinds of classical conditioning (e.g., trace conditioning). However, this also does not help much to draw the distinction between species. There are no significant differences in the learning abilities between mammal species, and most likely all vertebrates and even some invertebrates would still fall within the same category in terms of their “capacity to experience”.²

On the other side of the sentience coin, we find the concern that some species might have a higher capacity to experience. In a position paper, a European Science Foundation working group argued that nonhuman primates (NHPs) have a greater potential for suffering since they are : “distinguished by the very advanced nature of their social, cognitive, sensory and motor functions” (ESF 2009). A reasonable interpretation of this is that NHPs will be more harmed by research than other laboratory animals. But in which way?

We have analyzed this question in some detail elsewhere (Olsson and Sandøe 2010). In summary, in terms of capacity for sentience, it is unclear how most NHPs are different from other mammals which also share the capacity for experiencing pain and distress. Capacity for self-awareness may affect potential for suffering, but reasonable evidence to attribute this capacity only exists for great apes. The biological difference with clearest welfare relevance between NHPs and other mammal species used in research seems to be that primate species are not fully domesticated, making it more challenging to meet their needs in captive housing. On the other hand, there are also aspects in which primates may be better off in research than, for example, rodents: primates are usually trained to collaborate rather than restrained, their greater similarity to human beings facilitates the recognition of signs of poor welfare and higher concern for their welfare might encourage scientists to be more careful in how primates are treated.

Very recently, working on an analogy with pediatric research ethics , Fenton (2014) advocated that the cognitive capacities of chimpanzees may allow them to dissent from participating in research. It is not consensual whether chimps fulfill the conditions set, including, for example, whether they are capable of planning the future—which is far from being consensual (e.g., Shettleworth 2010; Suddendorf et al. 2009). Furthermore, it remains to be seen how to allow such dissent to be expressed in practice in a meaningful way. However, if these hurdles were overcome, this may be an interesting approach to develop a research ethics for nonhuman primates which not only respects but actually relies on their cognitive capacities.

In summary, with present knowledge there is little support for establishing differences which can motivate species choice to be a useful measure to reduce animal harm. Instead, the differences that society and the research community tend to make between less and more ethically problematic species are best understood in the light of the socio-zoological scale. This scale rates animals in terms of how greatly they are valued by humans, and places companion animal species and nonhuman primates at the top and rodents, fish, and invertebrates quite further down (Arluke and Sanders 1996). That the socio-zoological scale is based on what humans think about animals rather than on the characteristics of the animals themselves does not mean that it is ethically irrelevant. But, in our opinion, the difference between using a rhesus macaque or a fish in a given experiment is better described as more or less harmful to public sensitivity than more or less harmful to the animal.

We now turn to the other side of the harm–benefit equation, the benefits . Overall, research in neuroscience aims to deliver benefits for scientific knowledge and for human health and welfare, that is, benefits for humans. On one hand, basic research is conducted with the aim of understanding the functioning of the nervous system and the mechanisms underlying the diseases that affect it. On the other hand, applied research is carried out to develop treatments for such disorders.