A Review of the Institute of Medicine’s Analysis of using Chimpanzees in Biomedical Research

Notably, the NIH specifically requested an evaluation of the use of chimpanzees in predictive research, not basic research, an important distinction. Predictive research seeks to predict human response, while basic research seeks new knowledge regarding the material universe irrespective of whether that knowledge ever leads to practical advances (Greek and Greek 2010).For example, the determination of drug safety, toxicity, or efficacy, as well as the pathophysiology and natural history of disease, are a few of the ways that animals are used to model humans inpredictive research, whereas the use of animals such as fruit flies to search for the functions of genes in the fruit flies is referred to as basic research. Usually when the topic of the ethics and efficacy of the use of animals in biomedical research is discussed, the focus is often on predictive research which claims explicitly that animal models can predict human response to drugs and disease (Shanks and Greek 2009; Shanks et al. 2009).The request from the NIH is typical in that the content revolves around the use of chimpanzees for predictive purposes, focusing specifically on the necessity and efficacy as predictive models of human disease.

The committee concludes that “[w]hile the chimpanzee has been a valuable animal model in past research, most current use of chimpanzees for biomedical research is unnecessary, based on the criteria established by the committee.” (Institute of Medicine 2011, 4) Further, the committee makes the following two recommendations:

In the next section, we closely examine this concept of “genetic proximity to humans,” a crucial justification of the IOM’s recommendation. We approach this from an evolutionary biology perspective and consider what this proximity implies as well as what it does not imply. We argue that this proximity means different things depending the level of examination. We continue this theme in light of the fact that animals are examples of complex systems and as such are subject to specific characteristics that influence what can be expected in terms of inter-species extrapolation.

At the level of the periodic table, all life is similar, indeed in some sense, identical. That all humans are composed of carbon, hydrogen, oxygen, etc., is informative and can be quite useful in various domains of inquiry. However, it should be obvious that at this level of examination, using elemental similarity to determine how a patient will react to a drug or whether a particular animal species is sentientis not very useful. Similarly, examining life at the phylogenic level (e.g., noting that members of the Animalia and Plantae kingdoms are similar in that they are both all alive) takes place at a level of examination that is not helpful for predicting things like response to drugs and disease. For example, all members of the phylum Chordata have a heart that pumps blood to the tissues. As heart disease is one of the leading causes of morbidity and mortality among humans, one might expect a study of any given mammal to inform scientists about heart disease. However, such is not the case.(van der Worp and Macleod 2011; Fedorov et al. 2011; Greener et al. 2011; Gross 1985; Duff and McMillan 1949; Howard et al. 1972; Peters and Van_Slyke 1948 pp. 484, 500–501, 534–536; Shannon 1959 p 609]. Likewise, cancer affects humans and animals but the mechanisms and pathophysiology vary greatly. (Johnson et al. 2001; Leaf 2004; Brennan et al. 2010; Zielinska 2010; Editorial 2011; Brower 2011; Brody 2011; Rangarajan and Weinberg 2003).

An important and useful distinction can be made at this point regarding levels of examination. Levels of examination can be classified as either gross or fine.¹ On the gross level, mammals have hearts that pump blood. However, as our analysis of hearts moves to a fine level, we find salient differences among mammals, specifically between humans and other mammals. For example, humans are unique in that we respond to certain genetic and environmental factors by depositing plaque in our coronary arteries. If a scientist sought to learn about coronary artery disease in humans by studying mice, he would be disappointed as most species are not susceptible to coronary artery disease (see references just above). Likewise, at the gross level, the human immunodeficiency virus (HIV) can infect humans and chimpanzees. However, at finer levels of examination we notice that the results of infection with HIV are different: humans die from HIV, while chimpanzees essentially come down with a cold-like illness for a few days. There are myriad differences in immune response that account for this despite the fact that on the gross level both species have immune systems that have much in common (Gardner and Luciw 1989; Ferrari et al. 1993; Fultz 1993; Johnston 2000).

The brain is another excellent example of similarities on the gross level but where dramatic differences become apparent with a finer level of examination. Brains in mammals are composed of essentially the same kinds of cells and the gross structure is laid out similarly among species. In all mammals, there exists an area that controls motor activity, an area responsible for sensation, balance, hearing and so forth. But the response of the brain to drugs and disease differs significantly (Barnes and Hayes 2002; O’Collins et al. 2006; Schnabel 2008; Enna and Williams 2009; Geerts 2009; Regenberg et al. 2009; Mogil 2009; Unknown 2010; McArthur 2011).²

Genetic similarity is used to justify an expectation of similar responses to drugs and disease and therefore for using certain animals to test drugs and explore mechanisms of disease. For example, chimpanzees are nearly 99 % identical to humans in terms of nucleotide sequences. However, with regard to testing, this similarity is meaningless as nucleotide sequence is but one factor that determines what function the gene actually has and when and for how long it carries out this function. A gene must be placed in the context of other genes, gene networks, regulatory genes, modifier genes, and proteins before a full description can be established. This finer level of examination is necessary for discovering useful information regarding drug and disease response.³

Thus, at the fine level, making claims based on “genetic proximity” is quite often as meaningless as pointing to similarities in anatomy and physiology and assuming similarity in structure translates to similarity in origin. The importance of the similarity is dependent upon the question being asked and the level of examination involved. Thus, the level of examination (or organization) is important.

Futuyma famously notes that “[e]volution…is the central unifying concept of Biology. By extension, it affects almost all other fields of knowledge and must be considered one of the most influential concepts in Western thought” (Futuyma 1998). While this is generally accepted among scientists, such was not always the case. Darwin and Wallace’s idea fought for acceptance for decades. The early physiologists in France (including Claude Bernard, the father of our current paradigm of the use of animal models in drug and disease research) were among those that thought natural selection and descent with modification was simply wrong (Elliot 1987; Bernard 1957 (1865); LaFollette and Shanks 1994). Their belief, that all component parts in mammals were identical once size considerations were taken into account (Bernard 1957 (1865)), persists to this day.

However, even when the underlying assumptions are incorrect, advances in science can take place. For example, early morphologists like Cuvier rejected evolution in thinking that each category of animals was completely separate from all the others. Nonetheless, these researchers made significant pioneering advancements in science and made discoveries regarding phylogeny that, eventually, were interpreted as supporting evolution (Mayr 2002, 25). Importantly however, there are several differences between the early study of morphology and current study of human disease and drug reaction.

First, very little was known about phylogeny during Cuvier’s time whereas today much of what can be learned about the gross similarities among animals and humans has been uncovered. Despite its somewhat tautological nature this is not an insignificant point. In any field involving comparative anatomy and physiology the gross commonalities eventually are exhausted and the research modality must shift if more is to be learned in the most efficient way possible.

A second difference between the morphologists who denied evolution yet advanced science and today’s research is the level of examination. We now understand that drugs and disease act at the level of the gene or cell, not the level of the gross organ (although the effects are seen there). Hence, very small differences at the level of the gene negate the gross similarities in terms of drug development and disease research. Early anatomists and physiologists studied at the gross level where much similarity exists among species. The level of examination is still important in assigning gross characteristics. For example, in the 17th and 18th centuries, some human groups were not considered fully sentient beings just as many today do not consider animals sentient. A superficial examination of the brain and behavior refutes this notion. Thus, both the gross and fine level of examination are important depending on what question is being asked.

It is certainly true that from an evolutionary standpoint, we expect there to be fewer differences between humans and chimpanzees than between humans and mice or humans and yeast. Nevertheless, because humans and our closest phylogenetic relatives are complex adaptive systems, we should expect small differences to be of great biomedical significance. This problem is compounded the further away one moves in terms of a common evolutionary ancestor. If chimpanzees cannot predict human response to drugs and disease to the precision necessary in medical science, it is even less likely that mice, even genetically modified mice, will fulfill this role.

Some biomedical differences among species can be explained simply by differences in anatomy, regardless of evolutionary history. Examples of these differences between humans and NHHs would include sinusitis, infection of air sacs (which humans do not have) in NHHs, sleep apnea, musculoskeletal disorders of the back, a larger head and smaller pelvic outlet present difficulties in childbirth for humans, sudden infant death syndrome (SIDS), varicose veins, acne, hemorrhoids, slower wound healing in humans, and inguinal hernias. Differences that cannot be explained by anatomy include various disease of the heart, sexually transmitted diseases, various neurological diseases, and various infectious diseases. In particular chimpanzees infected with HBV or HCV rarely develop hepatocellular carcinoma or chronic hepatitis (Varki et al. 2011; Walker 1997; Gagneux and Muchmore 2004; Bettauer 2010). (See Tables 2 and 3.)

One reason humans differ from NHHs involves sialic acid. (See Tables 2 and 3) Sialic acid is a family of sugars that have a 9-carbon backbone. They are found at the end of glycan chains that are located on the surface of cells. There are fewer than 60 genes involved in sialic acid synthesis. Ten of these genes are hot spots in evolution demonstrating significant differences between humans and NHHs (Varki et al. 2011).

The evolution of new species must be placed into the context of complexity science. One reason why small differences in evolutionary history can result in profound differences in outcomes to perturbations is because animals and humans are complex rather than simple systems. A simple system can usually be defined and studied using Newtonian physics and can be expected to demonstrate a linear, or 1:1, response to perturbations. Simple systems are usually merely a sum of their parts and hence are also amenable to study by reductionism and determinism. Complex systems have very different characteristics (Csete and Doyle 2002; Sole and Goodwin 2002; Kitano 2002a, 2002b; Kauffman 1993; Ottino 2004; Alm and Arkin 2003; Goodwin 2001; Van Regenmortel 2002b, 2002a, 2004b, 2004a).

Complex systems are quite dependent upon initial conditions, as are chaotic systems. This fact is often overlooked for complex systems because the initial conditions for a complex system (such as, for example, financial markets), have been superseded by the outcomes of perturbations over the days and decades it has been in existence. In order to evaluate how the New York Stock Exchange will react to news of higher unemployment, for example, one does not need to understand the initial conditions of the NYSE when it was founded. Further, if one wishes to predict whether the Japanese Stock Exchange will respond to the same perturbation in the same fashion as the NYSE, one does not need to understand the initial conditions upon which it was founded. At the level of organization that is under study in such situations, the very distant past of the market is unimportant. However, if one wishes to predict the outcome to a perturbation for one living complex system, say a human, by using a second living complex system, say a mouse, then the initial conditions in the form of genetic makeup—including all the interactions, modifier genes, environmental influences, gene networks, and so on—must be considered. One very small difference in genetic makeup between the mouse and human can result in opposite reactions to a perturbation. In addition, complex systems are best-described using partial differential equations and all the values for the equations are not known. Thus predicting human response to drugs and disease based on mouse studies is unlikely in principle.

Complex systems are more than a sum of their parts; therefore reductionism has limits for evaluating a complex system. Another reason complex systems need to be studied as a whole is that they demonstrate emergent phenomena. Emergence is the appearance of a new trait or characteristic in a system that could not have been forecast even with complete knowledge of the component parts of the system. The level of examination is also important. Complex systems exhibit a hierarchy of organization and perturbations can result in opposite outcomes at different levels. The presence of feedback loops, environmental factors, redundancy, and robustness can also result in very different outcomes between two otherwise very similar complex systems.

Therefore, different species should be expected to react differently to the same perturbation to the system and this is has been confirmed empirically (LaFollette and Shanks 1995; LaFollette and Shanks 1996; Shanks and Greek 2009; Shanks et al. 2009; Sharp and Langer 2011; Chapman 2011; Giri and Bader 2011; Collins 2011; Caponigro and Sellers 2011; Leaf 2004; Ellis and Fidler 2010; Gura 1997; Sarkar 2009; American Paraplegia Society 1988; Littman and Williams 2005; Wall and Shani 2008; Smith et al. 1965; Smith and Caldwell 1977; Fletcher 1978) (Suter 1990 p 73; Lumley 1990 pp. 49–56; Heywood 1990 pp. 57–67). Moreover, merely changing a gene, either in the form of adding a human gene to a mouse or knocking out a gene in a mouse, should not be expected to substantially increase the predictive value of mouse models. This has also been confirmed empirically (LeCouter et al. 1998; Zutphen 2000; Morange 2001; Pearson 2002; Nijhout 2003; Van Regenmortel 2004b; Darlison et al. 2005; Shapiro 2007; Kieburtz and Olanow 2007; Young 2008; Enna and Williams 2009; Geerts 2009).

Given our discussion thus far, it should be clear that despite the IOM recommendation, even a species as closely related to humans as the chimpanzee will not be able to reliably predict human response to drugs and disease. Comparative genomics has confirmed the existence of very small differences between closely related species. Empirical evidence also suggests that predicting response to drugs and disease is nearly impossible and this empirical evidence is supported and explained by evolution theory and complexity theory. It should be evident that the IOM recommendation ignores the best science currently available in the form of evolutionary biology, evolutionary and developmental biology, complexity science, and empirical evidence.

In terms of science, what the committee could have addressed is the successful use of chimpanzees as heuristic models, as a source of tissues for use in humans, as bioreactors, as modes for other chimpanzees, for example to test vaccines designed for chimpanzees in the wild, and other scientifically viable uses of chimpanzees. However, the committee was specifically asked to consider the use of chimpanzees as predictive models and in this area chimpanzees as models have a history of failure. As we hope to have made clear, this failure is explained by the theory of evolution and complexity theory. The fact that the committee advocates such a conclusion, and similar conclusions, casts serious doubt on the scientific validity of the report.

It is difficult to remain consistent in one’s philosophy of science when that philosophy is not founded upon strong science. Without a strong appreciation for and adherence to evolutionary biology, avoiding claims based on similarity proves impossible for the committee.

The committee appears confused once again when it states that “[c]ontinued advances over the past decade in imaging, genetics, in vitro, and in silico models, and sophisticated rodent disease models have provided scientists with more tools that could be used in place of the chimpanzee” (Institute of Medicine 2011 p 29). First, rodent models cannot predict human response to drugs and disease any better than chimpanzee models (Greek and Greek 2010; Greek et al. 2012; Shanks and Greek 2009). Both fail as predictive models for human response to drugs and disease for the reasons outlined above. The committee frequently makes statements claiming that other animals can be used as replacements for chimpanzees because they function as well as chimpanzees. This is dubious as no species predicts human response to drugs and disease in humans.

Even other humans fail in this regard hence the current emphasis on personalized medicine (Greek et al. 2012). The evidence for this is quite convincing. Men differ from women in their response to drugs and disease (Holden 2005; Kaiser 2005; Klein and Huber 2010; Simon 2005; Wald and Wu 2010; Willyard 2009) and variation is also seen among ethnic groups (Cheung et al. 1997; Couzin 2007; Gregor and Joffe 1978; Haiman et al. 2006; Kalow 1991; Kopp et al. 2011; Spielman et al. 2007; Stamer and Stuber 2007; Wilke and Dolan 2011). No animal model more closely resembles a human than one monozygotic twin resembles another, yet even monozygotic twins vary in their response to drugs (Bruder et al. 2008; Dempster et al. 2011; Fraga et al. 2005; Javierre et al. 2010; Wong et al. 2005). Human variation in response to drugs and disease should inform society and the committee regarding the use of animal models to evaluate perturbations to the human system at a fine level of examination. Moreover, since individual humans react so differently to drugs and manifest different aspects of disease, the future of drug development will resemble figure 3 (Jørgensen 2011). This will not be possible using animal models.

Further, in vitro and in silico also fail in most cases, explaining why the failure rate for drugs in development is so high. Contrary to the committee’s statements that drug development is dependent upon animal models such as chimpanzees and other NHPs, the pharmaceutical industry as well as scientists in fields related to drug development, have opined that the animal model in general is a failure. This is born out by empirical evidence and failure rates (LaFollette and Shanks 1995; LaFollette and Shanks 1996; Shanks and Greek 2009; Shanks et al. 2009; Sharp and Langer 2011; Chapman 2011; Giri and Bader 2011; Collins 2011; Caponigro and Sellers 2011; Leaf 2004; Ellis and Fidler 2010; Gura 1997; Sarkar 2009; American Paraplegia Society 1988; Littman and Williams 2005; Wall and Shani 2008; Smith et al. 1965; Smith and Caldwell 1977; Fletcher 1978) (Suter 1990 p 73; Lumley 1990 49–56; Heywood 1990 pp. 57–67). In 2006, then US Secretary of Health and Human Services, Mike Leavitt stated, “[c]urrently, nine out of ten experimental drugs fail in clinical studies because we cannot accurately predict how they will behave in people based on laboratory and animal studies” (FDA 2006). The current focus in drug development is on developing ethical human-based testing and implementing it early in the development process (Kola and Landis 2004; Horrobin 2003; Seligmann 2004/5; Cressey 2011).

Thus, though the report claims to be primarily about science, its central focus is actually on the ethical aspects of the use of chimpanzees in biomedical research.

Rachels (1991) argues that when we take seriously a rejection of the view that humans are fundamentally and categorically different from other animals, we must also take seriously the implications of such a rejection as nothing less than a fundamental shift in the very project of ethics. Though acknowledging the fact/value gap,⁴ Rachels argues that even if there is no deduction of moral conclusions from factual premises, the fundamental assumptions of traditional morality are no longer tenable in light of evolutionary theory.

Specifically, since traditional morality is based on the assumption that species boundaries mark essential, hierarchical differences (with Homo sapiens at the top) and since evolutionary theory views species non-hierarchically with the notion of species itself being merely a pragmatic boundary, then the kind of categorical differences in moral worth and moral treatment foundational to traditional morality cannot be sustained. Whereas other philosophers writing on the moral status of animals argue that discrimination on the basis of species (i.e., speciesism) is morally arbitrary since species differences pick out no genuine morally relevant differences, Rachels goes further, claiming that if the species boundary marks no real boundary at all, then traditional morality’s correlation of such a boundary with moral value and significance is undermined from the start.⁵ If, as Mayr makes clear, thinking of organisms in typological terms is rendered obsolete by evolutionary theory, and since evolutionary biology gives primacy to the uniqueness of individuals within populations (Mayr 1994), then the idea that the moral status and treatment of individuals should be based on what is “normal” for their species should also be abandoned.

Rachels instead proposes an ethical framework he calls moral individualism, the view that “how an individual may be treated is determined, not by considering his group memberships, but by considering his own particular characteristics”⁶ (Rachels 1991, 173). Since humans and animals exhibit a complex pattern of similarities and differences, morality must respect this complexity. Morally relevant differences vary with the different kinds of treatment being considered. Thus, insofar as two individuals—regardless of their species—are similar, they should be treated similarly, while to the extent that they are different, they should be treated differently. On this view, “a difference between individuals that justifies one sort of difference in treatment might be completely irrelevant to justifying another difference in treatment” (Rachels and Rachels 2007, 24). So, for example, in some contexts, capacities such as intelligence and rationality are morally relevant properties (e.g., in deciding whether to admit someone to law school) and irrelevant in others (e.g., the treatment by a physician of a broken arm). This leads to moral individualism’s central principle of equality: “Individuals are to be treated in the same way, unless there is a difference between them that justifies a difference in treatment” (Rachels 1991, 96). With regard to animals specifically, the principle implies that “[o]ur treatment of humans and other animals should be sensitive to the pattern of similarities and differences that exist between them. When there is a difference that justifies treating them differently, we may; but when there is no such difference, we may not” (Rachels 1991, 197).

According to moral individualism then, it is not enough to simply assume (as does the committee) that there exist enough morally relevant differences between chimpanzees and mice such that mice can be used—where scientifically appropriate—as replacements for chimpanzees. For at the gross level, chimpanzees and mice are similar with regard to one morally relevant, salient property, namely, sentience, and it is this similarity in capacity that causes any such out-of-hand proclamations about the scientific suitability—and thus, the ethical significance—of the substitution of mice for chimps as ethically problematic.