Ethics of Research involving animals
Extrapolating the results of animal studies to humans: the scientific validity of animal research - continuation I
All modelling approaches face limitations concerning transferability and predictability
10.33 Given the vast complexity and variability of biological systems, it is not surprising that there are sometimes problems in developing effective experimental approaches in biomedical research and in extrapolating from model systems to humans (see paragraph 8.37–8.40). The difficulties, however, are an intrinsic part of any modelling approach that relies on surrogates for the range of organisms of interest. Nor are they confined to animal studies, but are also encountered in developing and applying other experimental approaches, such as in vitro and clinical studies. None of these methods can reproduce exhaustively all the features that characterise the wide diversity and variation of genetic and biological processes that occur in a population of humans, as is clear from the following examples:
i) Limitations of in vitro research: differences between human cells in vitro and in vivo can pose challenges in extrapolating findings from research on the functioning of human cells in culture to the functioning of human cells in vivo (see Chapter 11 for further discussion).9 Yet more acute challenges arise in using the findings from cell culture studies to make predictions relating to the integrated physiology of intact tissues, organs or the whole human body.
ii) Limitations of human clinical trials: even if the animal-research stage was omitted from the development of new medicines, intrinsic problems resulting from the way clinical trials are conducted remain. First, human clinical trials typically involve testing a drug on 1,000–5,000 human volunteers and patients. If a side effect occurs in 1 in 10,000 patients, it is likely to become apparent only after the product is marketed (see Boxes 8.6 and 8.7). Secondly, human trials usually involve a relatively homogeneous sample of patients in order to distinguish clearly between the effects of the therapy (the ‘signal’) against the background of variation between different patient’ responses (the ‘noise’).10 Such trials, moreover, frequently provide little, if any, information about the effects of drug interactions, since they usually do not mimic the actual situation in which patients may take several different medicines at the same time.11 Uncertainties about the effects of treatments in the clinical setting are therefore inevitable,12 and clinicians must exercise judgement in extrapolating the results of clinical trials to individual patients (see paragraph 11.21).13
Box 10.1 : Toxicity studies in humans:
10.34 These observations help to explain why adverse reactions sometimes occur in humans when medicines are brought to the market after testing in vitro, in animal studies and in human clinical trials, none of which individually, or collectively, have allowed the prediction of these effects. Nevertheless, such adverse reactions generally occur in relatively few patients, and only a small fraction of marketed medicines have been withdrawn for safety reasons (Boxes 8.6 and 8.7).
10.35 To what precise degree animals can be said to be useful models of human disease continues to be controversial. Taking into account evidence presented in Chapters 5–9 and the above discussion, we note that there have been a great number of cases where animals have been used successfully to provide models for humans (or other animals of different species) We therefore agree with the finding of a recent Report by the Animal Procedures Committee (APC), which observed that: ‘the scientific validity of animal experiments is a condition capable of being fulfilled, but has to be judged case by case and subjected to detailed critical evaluation’.14
10.36 We draw a similar conclusion with regard to the assertions that animal experiments lack internal validity because they sometimes fail as a result of poor experimental design or other methodological problems. While it is clear that such examples exist (see paragraphs 6.32, 6.37 and Box 8.4), they are insufficient to support the claim of a general flaw. Rather, those advocating the use of animals in research take the view that these cases point to a need to carry out a critical evaluation of any design of a study, regardless of the method or subject employed (be it computer studies, in vitro, animal or human).15 With regard to the special case of thalidomide, critical reflection helped prompt the introduction of regulations that require more rigorous and consistent testing of medicines in animals in order to help prevent further tragedy (Box 8.4) next page
Footnotes9 This point draws on Horrobin’s provocative discussion in a recent opinion: Horrobin DF (2003) Modern biomedical research:
an internally self-consistent universe with little contact with medical reality? Nat Rev Drug Disc 2: 151–4.
10 Fletcher RH (2002) Evaluation of interventions J Clin Epidemiol 55: 1183–90.
11 Stricker BHCh and Psaty BM (2004) Education and debate article: detection, verification and quantification of adverse drug
reactions BMJ 329: 44–7.
12 Chalmers I (2004) Editorial: Well informed uncertainties about the effects of treatments: how should clinicians and patients
respond? BMJ 328: 475–6.
13 Fletcher RH (2002) Evaluation of interventions J Clin Epidemiol 55: 1183–90.
14 Animal Procedures Committee (2003) Review of the cost-benefit assessment in the use of animals in research (London: HO), p26
15 see Chapter 6, footnote 40.