These consider some of the issues relating to the use of animals in biomedical research:
There have been very few reliable, systematic evaluations of the human relevance of animal research (Pound et al 2004). Many of the proponents of animal research cite single, selected studies as evidence that animal research ‘works’ in a particular area, but single studies – which may be flawed or biased – do not constitute sufficient, high quality evidence. The best evidence comes from systematic reviews. The systematic review method involves taking all the relevant studies in a field of interest and synthesising their findings in order to produce a definitive answer. However, relatively few systematic reviews of animal studies have been conducted and it is rare for them to explore issues relating to human relevance.
An exception is provided by Perel and colleagues (2007), who conducted systematic reviews of animal studies for six different conditions and compared them with systematic reviews of human studies for the same six conditions. They found that the animal and human studies agreed in three cases and disagreed in three cases. They concluded that the lack of agreement may have been due to either the inability of animal models to accurately mimic human diseases, or to the poor quality of the animal studies. Another systematic review found that only a third of animal study findings agreed with the findings of the corresponding human randomised trials (Hackam and Redelmeier 2006).
In 2016, the UK government’s chief scientific advisor suggested that the time had come to evaluate the human benefits of animal research. In a lecture to the animal research community, he asked: ‘To what extent have we as a community, ever subjected our claims about how vital animal research has been to human health to the same level of scrutiny we’d apply to those claiming to have discovered a new cure? And I think if not, we must.’ (Walport 2016)
It is estimated that between only 5% (Contopoulos-Ioannidis et al 2003) and 10% of animal study interventions result in approved use in humans (Kola and Landis 2004; van der Worp et al 2010). However the rate varies widely by disease; cancer is estimated to have a 5% success rate (Kola and Landis 2004), Alzheimer’s disease a 0.4% success rate (Cummings et al 2014) and stroke a dismal success rate of 0.1% (Howells et al 2012). The highest rate -18% – appears to be for cardiovascular disease (Vatner 2016).
Elsewhere there is a similar story. In the field of spinal cord injury, none of the 22 drugs that worked in animals turned out to work in humans (Geerts 2009). In the case of inflammatory diseases, there is almost no correlation between human and mouse data (Seok et al 2013). Every approach to treating sepsis that was successful in animals has failed in humans (Leist and Hartung 2013). All but one of the experimental treatments that improved motor neurone disease in animals failed in human trials, and the benefits of the one successful treatment, in terms of extended survival, are considered negligible (Perrin 2014).
Certainly it is common to hear optimistic reports about medical breakthroughs on the news. The usual story is that a promising treatment for a certain condition has been found in mice and hope is expressed that this could eventually lead to a treatment for the human condition. However such news items are never accompanied by information on the very poor odds of this happening, or the difficulties involved in extrapolating such findings to humans (Woloshin et al 2009).
Scientists who issue this type of news release are able to hold out the promise of animal research without actually having to deliver anything at the time. It is unlikely that anyone would follow up the study in twenty years’ time to see whether it did actually produce any benefits for humans. However one group of researchers set out to do something like this. They found that of 101 research papers that clearly stated a promising finding, some twenty year later only one had led to a treatment that was subsequently used extensively in humans (Contopoulos-Ioannidis et al 2003).
Unfortunately not. There is good evidence that only a few animal researchers are aware of the many ways in which their experiments can be biased (Reichlin et al 2016) and that only a few researchers take steps to minimise the possibility of these biases (Kilkenny et al 2009). For example only a minority of animal researchers allocate animals to study groups randomly (Hirst et al 2014; Macleod et al 2015; Henderson et al 2015) with the result that unconscious bias creeps in. Studies that do not use random allocation are more likely to report positive findings (Bebarta et al 2003; Hirst et al 2014). Another way in which bias can be introduced is when the person assessing the outcomes of an experiment is aware which group received the experimental treatment. This is a problem in animal research (Henderson et al 2015; Bebarta et al 2003; Perel et al 2007; Kilkenny et al 2009; Hirst et al 2014; Macleod et al 2015); studies in which the researcher assessing outcomes knew which group received the experimental treatment were more likely to report positive findings (Bebarta et al 2003; Vesterinen et al 2010).
Furthermore, simplistic statistical analyses are often used that do not account for factors such as the age or sex of the animals. Not taking such factors into account may alter the results of experiments. For example there are at least 50 publications describing drugs that extend the lives of mice with motor neurone disease. Researchers identified factors (such as age and sex) that might have altered this body of research and then repeated the experiments, this time correctly taking account of such factors. They found that none of the drugs now extended the mice’s lives and concluded that the original positive findings were unlikely to have been due to the effect of the experimental drugs, but were most likely due to the experiments not being conducted properly (Scott et al 2008).
As a result of the lack of scientific rigour in animal research, it can be seen that large bodies of animal research report positive findings that are not actually real, but that are a function of poorly designed experiments. At present most animal research is of such poor quality that no reliable conclusions may be drawn from it. This wastes animals’ lives, squanders research funding and endangers humans who participate in clinical trials that are based on misleading animal data.
In general, no. The reporting of research findings should be complete and unbiased, i.e. it should include all the findings, not just those with positive results. Unfortunately however, there is evidence that some researchers select, from among all the analyses performed, only those with the best results. This practice can lead to the benefits of drugs that are tested on animals being overestimated (Tsilidis et al 2013).
Publication bias is a term that describes the phenomenon whereby studies are more likely to be published if they report positive findings. This means that negative findings – for example where an experimental drug has not been found to work in animals – tend not to be published. This is a significant issue in animal research (Korevaar et al 2011; Mueller et al 2014). The problem is that if only positive findings are published, the benefits of that body of research are overestimated, as has happened in the field of animal studies of stroke (Sena et al 2010). This means that overoptimistic conclusions are drawn about drugs that are tested in animals. However when these drugs are trialled in humans – possibly exposing those humans to harm – they ultimately fail, wasting considerable time and funds.
Another consequence of negative findings not being published is that other researchers may repeat the failed experiments because they were unaware of the results, meaning that more animals’ lives and funding are wasted.
When researchers administer a substance to animals, they get plenty of feedback on the effectiveness of that substance in the animals tested. However, results nearly always differ between species, so testing a substance on animals is not a reliable method of predicting the human responses to that substance. Some drugs with potential benefit for humans might not be approved because they have been found to harm animals; conversely, drugs found to be safe in animals might then go on to harm humans.
It is estimated that more than 10,000 people are killed every year in the UK by adverse reactions to prescription medicines (Pirmohamed et al 2004). In the US, adverse drug reactions are the fourth –sixth leading cause of death (Lazarou et al 1998). The arthritis drug Vioxx caused thousands of deaths prior to being taken off the market in 2004. It is difficult to estimate the exact number of deaths, but a scientist at the United States Food and Drugs Agency (FDA) calculated that approximately 55,600 people may have died as a result of Vioxx in the US alone (www.whistleblower.org/dr-david-grahams-full-story). While the FDA and the company that developed Vioxx failed in their responsibilities (Topol 2004), animal tests failed to predict that Vioxx would cause deaths and adverse reactions in humans.
Evolutionary biologists (e.g. Perlman 2016) note that while animal research might be useful for understanding processes that arose early in evolution (i.e. processes that humans share with other species) it is less useful for understanding the chronic diseases suffered by modern humans because these chronic diseases are a result of human lifestyles and the unique way in which human life has evolved. Although human and non-human animals may have many genetic, biochemical and physiological similarities, when it comes to complex and evolved living systems even minor differences can result in significant differences in biological processes and outcomes.
Phase 1 trials (where drugs are first tested in human volunteers) have shown that even the smallest biological differences between humans and animals can lead to disaster when first applying animal data to humans. TGN1412 was tested in non-human primates (NHPs) because of their close relationship to humans but soon after being given a dose 500 times smaller than that found safe in animal studies, all six human volunteers began to experience a chain of events which rapidly led to severe inflammation and multiple organ failure (Attarwala 2010). Intractable differences between species mean that animals cannot reliably predict how the human body will respond to a disease or a drug.
Many believe that NHPs, because of their close relation to human primates, must have relevance for human medicine. However the ability of NHPs to predict human adverse drug reactions is poor and some drugs that were safe for NHPs have gone on to injure or kill humans. For example the arthritis drug Vioxx, which killed approximately 55,000 people in the US alone, was found to be safe in monkeys. NHPs are also used for brain research, yet the most dramatic differences between humans and other primates are found in the brain. Perhaps this helps explain why, of over one thousand drugs for stroke that have been developed and tested in primates and other animals (O’Collins et al 2006), all but one have failed and even harmed patients in clinical trials (e.g. Tirilazad International Steering Committee 2001). And even the benefits of the one ‘successful’ drug are controversial (Sandercock and Ricci 2017).
Animal cancer is not the same as human cancer and substances known to be carcinogenic in one species are not necessarily carcinogenic for other species. Animal models are unable to mimic the complex process of human carcinogenesis, physiology and progression, with the result that they rarely translate into benefits for humans with cancer.
The average rate of translation from animal models to successful clinical cancer trials is less than 8%. Experts in the field are increasingly recommending a reduction in reliance on animal models and the use of new (non-animal) approaches in order to transform cancer drug discovery (Jackson and Thomas 2017), including microdosing in (Phase 0) human trials, in vitro human cell-based assays, in silico (computer) models, increased emphasis on epidemiology (Mak et al 2014), and the use of human tissue models (Jackson and Thomas 2017).
Dr Azra Raza, Professor of Medicine at Columbia University in New York, and Science Adviser to Safer Medicines Trust gives a powerful and moving 13-minute TEDx Talk on the need for human-focused cancer research.
Agents that are teratogenic (cause birth defects) to some species may have little or no teratogenic effect in others. For example, after women who had taken thalidomide for ‘morning sickness’ started giving birth to babies with malformed limbs, researchers tested thalidomide in animals to see if they produced similar malformations. Thalidomide was tested in numerous animal species without producing any birth defects (Taussig 1962). As Schardein (1985) writes: ‘In approximately 10 strains of rats, 15 strains of mice, 11 breeds of rabbits, 2 breeds of dogs, 3 strains of hamsters, 8 species of primates and in other such varied species as cats, armadillos, guinea pigs, swine and ferrets in which thalidomide has been tested, teratogenic effects have been induced only occasionally.’
There is some uncertainty and controversy about whether thalidomide was properly tested on pregnant animals before humans. However, even if the correct tests had been conducted it is unlikely they would have prevented the disaster. As Teeling-Smith (1980) notes: ‘With thalidomide (…) it is only possible to produce specific deformities in a very small number of species of animal. In this particular case, therefore, it is unlikely that specific tests in pregnant animals would have given the necessary warning: the right species would probably never have been used.’
When polio first appeared around 1835, it rapidly paralysed and killed its victims. In 1908, a virus was suspected and scientists began working on a vaccine. They discovered the polio virus in human intestines as early as 1912, suggesting entrance through the digestive tract. However, because animal research indicated that monkeys contract polio nasally, and because scientists working on the vaccine ignored the human findings in favour of the animal findings, the development of an effective vaccine was postponed for decades (Sabin 1965; Parish 1968). In 1941, after conducting human autopsies, Dr. Albert Sabin found the virus confined to the gastrointestinal tract as had been documented nearly 30 years earlier. He noted that ‘(…) prevention was long delayed by the erroneous conception of the nature of the human disease based on misleading experimental models of the disease in monkeys.’ (Sabin 1984)
Animal research also significantly side-tracked the development of penicillin. In 1929, Alexander Fleming observed penicillin as it killed bacteria in a Petri dish. Intrigued, he administered the compound to bacteria-infected rabbits, hoping it would do the same thing. Because the penicillin was ineffective against the rabbits’ infection, Fleming set aside the drug for another decade. Years later, he administered the drug in desperation to a dying patient, for whom all other treatments had been ineffectual. The penicillin saved his patient’s life and the rest is history. Fleming later commented, ‘How fortunate we didn’t have these animal tests in the 1940s, for penicillin would probably never been granted a license, and possibly the whole field of antibiotics might never have been realised’ (Parke 1994).
Advocates of animal research often cite the development of insulin as an example of the value of animal research. While it is true that insulin obtained from animals saved the lives of many diabetics, animal research was not useful in investigating the cause of this disease. Physicians in the late 18th century first linked diabetes to changes in the pancreas seen at human autopsy. Although dogs, cats, and pigs became diabetic when their pancreases were removed, their symptoms led researchers to conjecture that diabetes was a liver disease, throwing diabetes research off track for decades. In 1922, scientists spoke out against the animal experiments (Roberts 1922), pointing out that human autopsy had in fact shown the pancreas to be the vital organ in diabetes, and that it was in vitro research that was responsible for the isolation of insulin.
The discovery of the DNA double helix, arguably the 20th century’s most important medical breakthrough, was thanks to non-animal technology and in vitro research. Similarly, we owe our ability to view the heart’s blood vessels to German urologist and Nobel prize winner, Werner Theodor Otto Forssmann. He had tried his procedure on rabbits but they had died. Fortunately Forssman distrusted his animal experiments and in the tradition of many physician scientists, experimented on himself instead. By sliding a catheter into his own arm and taking x-rays, he showed that a catheter could be threaded to the heart without problem. As a result doctors can now monitor blood vessels, detect problems and implement preventive measures.
In 2004 Safer Medicines Trust commissioned a survey of 500 General Practitioners. The survey was conducted by global market research organisation TNS Healthcare who selected the GPs to ensure a thorough demographic and geographical UK spread. The results revealed a surprising level of distrust in the data obtained from animal research:
- 82% of GPs were concerned that animal data could be misleading when applied to humans
- Only 21% would have more confidence in animal tests for new drugs than in a battery of human-based safety tests
- 83% would support an independent scientific evaluation of the clinical relevance of animal research
Clearly, a silent majority of today’s doctors are aware that animal tests do not guarantee the safety and efficacy of new drugs.
Many factors perpetuate animal research, the most obvious of which is tradition. The animal research paradigm is very deeply embedded within research institutions. Laboratories, researchers’ careers, technologies, publication practices and funding streams are now ‘locked-in’ to the paradigm. For example, it is easier and faster to publish papers based on animal research than on human research, and a good publication record will attract funding to continue that research.
Pharmaceutical companies continue to do animal tests because regulators require them to do so and because (despite their lack of reliability) animal tests provide liability protection in court when drugs injure or kill people. Yet most of the animal tests accepted by regulators have never been validated.
The 3Rs (the reduction, refinement and replacement of animals in research) are principles that were developed over 50 years ago in order to encourage a more ethical approach to the use of laboratory animals. Specifically the aim of the 3Rs is to reduce the number of animals used in research, to refine experimental procedures to cause less suffering and to replace animal experiments with non-animal ‘alternatives’.
The 3Rs may improve the lives of some laboratory animals to some extent, which of course has merit. However there is concern that this ethical framework, developed in the 1950s, may no longer be fit for purpose (Ashall and Millar 2013) and there is also evidence that opportunities to implement the 3Rs may be missed (van Luijk et al 2013; Balcombe et al 2013). One commentator has suggested that the 3Rs function primarily to achieve consensus within a contested domain (Hobson-West 2009).
However a more fundamental problem with the 3Rs is that they are based on an assumption that animal research is scientifically valid and that it is therefore a ‘necessary evil’. As such, they have the effect of deflecting attention away from the central issue, which is doubt over the scientific validity of using animals as models of human disease.
- In vitro (test tube) research has been instrumental in many great scientific and medical advances. For example the discovery of antibiotics, the structure of DNA, and the development of many of the vaccines we use today.
- Epidemiology (research into the causes of health and illness in populations) has revealed that folic acid deficiency causes birth defects, tobacco smoking causes lung cancer, lead damages children’s brains and social inequality leads to ill-health.
- Post-mortem studies have been used for centuries to further understanding of the human body and have led to many clinical benefits, including advances in congenital heart disease. Post mortem studies remain an excellent way of studying the effects of a disease on the whole body.
- Clinical studies: well-designed clinical trials are able to establish whether current practice is actually the best option. Some practices have been shown in clinical trials to cause more harm than benefit, e.g. the use of corticosteroids after brain injury.
- Human tissue is vital in the study of human disease and drug testing (animal tissue differs from human tissue in crucial ways). Human tissue is used for a range of different purposes, including understanding disease progression, developing diagnostic and screening tests, and testing the effects of drugs. Tissue banks store human tissue for use in research, e.g. tissue from people with certain diseases, or from different parts of the body.
- Human stem cells: the generation of human induced pluripotent stem cells (iPSC) has been an important breakthrough. iPSC possess the genetic background of individuals so can be used to create disease- or patient-specific models, such as ‘organoids’.
- Organoids are simplified in vitro versions of organs, capable of modelling some specific function of that organ (Lancaster and Knoblich 2014).
- Organs on chips are micro physiological systems that enable basic biological processes to be studied and the effects of drugs to be investigated. By identifying safety and efficacy issues earlier in the drug development process they may enable the selection of drug candidates that are more likely to succeed in human clinical trials. The US is making significant investments in organ-on-chip technologies (Marshall et al 2018).
- In silico (computer) approaches and mathematical modelling are now being used to better understand the complex chain of processes that occur when a chemical enters the human body, allowing a better understanding of how toxicity is expressed in the body. In an in silico ‘drug trial’ the computer model predicted the risk of human drug-induced heart arrhythmias with 89% accuracy, compared with animal studies that showed up to 75% accuracy (Passini et al 2017). In another example, in which the world’s largest machine-readable toxicological database was used to predict the toxicity of new chemicals, it clearly outperformed animal tests (Luechtefeld et al 2018).
- Imaging technologies: magnetic resonance imaging (MRI), functional MRI (fMRI), positron emission tomography (PET) and other imaging technologies are now able to offer a view of the human body that cannot be gained by studying animals.
- Microdosing is a process whereby drugs are administered in doses small enough to be safe, but large enough for the cellular response to be studied, enabling potential new drugs to be safely tested in humans. All new drugs are eventually tested on humans but new microdosing techniques can be used to achieve this more safely.
- Post-marketing drug surveillance is the practice of monitoring drugs after they have been marketed. Potentially this can help to identify any unexpected adverse drug reactions much sooner.
- Prevention is always more effective than cure. It is estimated for example, that 90% of stroke risk factors are modifiable through lifestyle changes (O’Donnell et al 2016).
Replacing the animal model is not about finding a one-to-one ‘alternative’ to every procedure that involves animals, it is about a completely new approach to research that is committed to using the best and most physiologically relevant research methods for investigating human health and illness.