Until recently, critical evaluations of the accuracy of such claims have been rare. However, several large-scale systematic reviews of the value of the animal experiments have now been published in scientific and medical journals, by the speaker and his scientific colleagues. Several have received awards at international scientific conferences.
The outcomes have been consistent: animal experiments have contributed far less than advocates would have us believe.
This presentation summarises these recent results, and comprehensively reviews the alternatives to invasive animal use with biomedical research, toxicity testing, and education.
Animal Experimentation Scrutinised: Humane Teaching Methods and Computer Simulations
1. Animal Experimentation
Scrutinised
Andrew Knight BSc., BVMS, CertAW, MRCVS, FOCAE
Animal Consultants International
www.AnimalConsultants.org
2. 1997
“ What you’ve seen so far is only the tip of the iceberg
compared to what you will have to do to animals later in the
veterinary course. Perhaps you should re-think your choice
of career… ”
16. ‘Alternative’
veterinary surgical training
1. Knot-tying boards, plastic organs and similar
models: basic manual skills such as suturing and
instrument handling.
3. Ethically-sourced cadavers: simulated surgery.
5. Real patients: observing, assisting with, and then
performing beneficial surgery under close
supervision (e.g. shelter animal neutering programs).
20. Alternative Veterinary Surgical
Program, 2000
External clinical experience in private clinics or animal
shelters assisting with or participating in surgery and
anaesthesia.
Sterilisations of real patients, e.g., from animal shelters, at
Murdoch.
Attendance at all of the terminal surgical laboratories as
observers.
21. Simulated abdominal surgeries on a “DASIE” (Dog
Abdominal Surrogate for Instructional Exercises).
Ethically-sourced cadaver surgery: abdominal and
orthopaedic surgeries.
22. Outcomes
Jointly we did not participate as surgeon or assistant surgeon
in a total of at most 13 scheduled surgeries at Murdoch.
We performed or assisted with a total of at least 62
additional surgeries instead, not including the abdominal
surgeries I performed on a “DASIE” surgical simulator.
Surgeries performed under supervision, mostly in private
practice.
23. Depth: Jointly we sterilised 45 dogs and cats, including 21
spays.
Breadth: We also participated in a range of other surgeries as
well, e.g., umbilical hernia repair, cruciate ligament repair,
cutaneous polyp and lump excisions, aural haematoma
excision, abdominal surgeries (exploratory laparotomy,
enterotomy, enterectomy, partial splenectomy, spay),
orthopaedic surgeries (trochanteric osteotomy, stifle
arthrotomy).
Similar depth and breadth of anaesthetic experience.
24. The effectiveness of humane
teaching methods
Knight A. The effectiveness of
humane teaching methods in
veterinary education. ALTEX:
Altern Anim Experimentation
2007;24(2):91-109.
www.HumaneLearning.info,
‘Published papers, Comparative.’
25. Results
12 papers published from 1989 to 2006 described 11 distinct
studies of veterinary students:
9 assessed surgical training—historically the discipline
involving greatest harmful animal use.
Humane method
Superior Equivalent Inferior
45.5% (5/11) 45.5% (5/11) 9.1% (1/11)
26. All disciplines
Knight A, Balcombe J & De Boo J, www.HumaneLearning.info,
‘Published papers, comparative.’
At least 33 papers sourced from the biomedical and educational
literature, covering all educational levels and disciplines, describe
studies that have compared the ability of humane alternatives to
impart knowledge or clinical or surgical skills.
Humane method
Superior Equivalent Inferior
39.4% (13/33) 51.5% (17/33) 9.1% (3/33)
27. Conclusions
Well-designed humane alternatives usually perform at least
as well as methods that rely upon harmful animal use, in
some cases achieving superior learning outcomes.
Their financial and time savings, repeatability, increased
flexibility of use, and potential to increase active learning and
computer literacy, all provide other important advantages
when compared to traditional methods reliant upon harmful
animal use.
28. Alternatives sources
From Guinea Pig to Computer Mouse: Alternative Methods for
a Progressive, Humane Education. www.InterNICHE.org
www.vetmed.ucdavis.edu/Animal_Alternatives
www.clive.ed.ac.uk
Alternatives libraries, free on-line computer simulations,
comprehensive alternatives databases, academic reviews of
leading alternatives, and hundreds of educational studies of
alternatives organized by discipline:
www.HumaneLearning.info
www.EURCA.org
29. Other major student successes
2000 University of Sydney: all terminal veterinary surgical labs
stopped, alternatives introduced, conscientious objection policy
passed
2000, 2004 University of Illinois: all physiology vivisection labs
stopped, conscientious objection policies passed
2000 Massey University, New Zealand: most veterinary
physiology vivisection labs stopped, conscientious objection policy
passed
2004 University of Queensland: alternatives to terminal
veterinary surgical laboratories introduced
2004 University of Melbourne: alternatives to terminal
veterinary surgical laboratories introduced
32. Scientific resistance to alternatives
Non-compliance of US researchers with the alternatives
regulations of the Animal Welfare Act:
Most common: inadequate consideration of alternatives
(600 - 800 research facilities).
Fourth most common: unnecessary experimental
duplication (~ 250 facilities).
Others: inadequate justification for animal numbers,
alleged uncertainty of research personnel about signs
indicative of pain and/or distress (USDA-APHIS-AC,
2000).
33. Supporting animal experimentation
Animal experimentation is vital for preventing, curing or
alleviating human diseases (e.g. Brom 2002, Festing 2004).
The greatest achievements of medicine have been possible
only due to the use of animals (e.g. Pawlik 1998).
The complexity of humans requires nothing less than the
complexity of laboratory animals to effectively model during
biomedical investigations (e.g. Kjellmer 2002).
Medical progress would be “severely maimed by prohibition
or severe curtailing of animal experiments,” and
“catastrophic consequences would ensue” (Osswald 1992).
34. Concordance or discordance?
Drugs causing serious side effects or death in animals
that are harmless to humans:
Penicillin
Morphine
Aspirin
…
35. Drugs released onto the market after passing
rigorous testing in animals, and very limited testing
in humans, that have caused serious human side
effects:
TGN1412 (UK, 2006)
Vioxx
Thalidomide, Eraldin, Chloramphenid, Ibufenac, Flosint,
Zipeprol, Zomax, Accutane, Benedectin, Phenformin
Many, many more…
Such adverse drug reactions are the 4-6th leading cause of
death in US hospitals, and kill over 10,000 people annually
in the UK.
36. Calls for systematic reviews
Pound and colleagues (British Medical Journal 2004):
Clinicians and the public often consider it axiomatic that
animal research has contributed to human clinical knowledge,
on the basis of anecdotal evidence or unsupported claims.
These constitute an inadequate form of evidence for such a
controversial area of research, particularly given increasing
competition for scarce research resources.
Hence, formal evaluation of existing and future animal
research is urgently required, e.g., via systematic reviews of
existing animal experiments.
37. Systematic reviews:
‘gold standard’ evidence
Critically examine human clinical or toxicological utility of
animal experiments
Examine large numbers of experiments
Experiments selected without bias, via randomisation or
similarly methodical and impartial means
Studies published in peer-reviewed biomedical journals
38. Literature survey 2007
27 systematic reviews of the utility of
animal experiments in advancing human
clinical outcomes (20), or in deriving
human toxicity classifications (7)
Three different approaches sought to
determine the maximum clinical utility
that may be achieved by animal
experiments …
39. 1. Experiments expected to lead to
medical advances
Lindl et al. (2005 & 2006) examined animal experiments
conducted at three German universities between 1991 and
1993, that had been approved by animal ethics committees at
least partly on the basis of claims by researchers that the
experiments might lead to concrete advances towards the cure
of human diseases.
For 17 experiments meeting the inclusion criteria, citations
were analysed for at least 12 years. 1,183 citations were
evident.
However …
40. Only 8.2% of all citations (97) were in clinical publications.
Of these, only 0.3% of all citations (4 publications)
demonstrated a direct correlation between the results of animal
experiments and human outcomes.
However, even in these four cases the hypotheses that had
been successfully verified in animals failed completely when
applied to humans.
None of these 17 experiments led to any new therapies, or,
indeed, any beneficial clinical impact during the period
studied!
41. 2. Clinical utility of
highly cited animal experiments
Highly cited animal experiments are most likely to be
subsequently tested in clinical trials. Hence, Hackam &
Redelmeier (2006) searched for experiments with more than
500 citations, published in the 7 leading scientific journals
when ranked by journal impact factor.
76 animal studies were located with a median citation count of
889 (range: 639-2,233).
However…
42. Only 36.8% (28/76) were replicated in human randomized
trials. 18.4% (14/76) were contradicted by randomized trials,
and 44.7% (34/76) had not translated to clinical trials.
Ultimately, only 10.5% (8/76) of these medical interventions
were subsequently approved for use in patients.
43. Translation rates of most animal
experiments are even lower
Most experiments are neither highly cited nor published in
leading journals. Many experiments are not published at all.
The selective focusing on positive animal data while ignoring
negative results (optimism bias) is one of several factors
identified that may have increased the likelihood of translation
beyond that scientifically merited.
Rigorous meta-analysis of all relevant animal experimental
data would probably significantly decrease the translation
rate to clinical trials (Hackam, 2007).
44. Poor methodological quality
Additionally, only 48.7% (37/76) of these highly cited animal
studies published in leading journals were of good
methodological quality.
Few included random allocation of animals, adjustment for
multiple hypothesis testing, or blinded assessment of
outcomes.
Accordingly, Hackam & Redelmeier cautioned patients and
physicians about extrapolating the findings of even highly
cited animal research to the care of human disease.
45. 3. Chimpanzee experimentation
Prominent advocates of invasive research on captive chimpanzees
such as Vandeberg et al. (Nature, 2005) have called passionately for
the funding of such research to be increased.
They have stated that such research has been of great importance
during our struggles against major human diseases such as AIDS,
hepatitis and cancer, and that the genetic similarities between
humans and chimpanzees — our closest living relatives — makes
them ideal biomedical research models.
46. Contributions to biomedical knowledge
749 studies of captive chimpanzees or chimpanzee tissues, conducted
from 1995-2004:
Figure 1: Chimpanzee experiments
2% 1995-2004 (total 749)
2%
3%
3% Biology (363)
Diseases: virology (311)
Therapeutic
48% investigations (26)
Diseases: parasitology
(23)
42% Miscellaneous (14)
Diseases: other (12)
48. Figure 3: Virology experiments
11% (311 of 749)
2% HCV (97)
2% HIV (97)
3% HBV (29)
31%
3% RSV (12)
HEV (11)
4%
STLV (9)
4%
HIV & SIV (8)
9%
SIV (7)
TTV (7)
31%
21 others (34)
HCV = hepatitis C v., HIV = human immunodeficiency v., HBV = hepatitis B
21 others: Six: FV. Four: HAV. Two each: GBV – B, v., RSV = respiratory syncytial v., HEV = hepatitis E v., STLV = simian T-
HIV & HV, IV, PIV, Noroviuses. One each: cell lymphotropic v., SIV = simian immunodeficiency v., TTV = transfusion-
transmitted v., FV = foamy v (human and simian FV), HAV = hepatitis A v.,
Bacteriophages, Dengue v., Ebola v., HCMV, HGV, GBV-B = GB virus B, HV = herpes v., IV = influenza v., PIV = parainfluenza
HMPV, H/S TLV, LCV, Papillomaviruses, RV2, v., HCMV = human cytomegalovirus, HGV = hepatitis G v., HMPV = human
Rhinovirus, VZV, WMHBV, Unspecified.
metapneumovirus, H/S TLV = human/simian T-cell leukemia v., LCV =
lymphocryptoviruses, RV2 = rhadinovirus (or gamma-2-herpesvirus)
genogroup 2, VZV = varicella-zoster v., WMHBV = woolly monkey hepatitis
B v.
49. Therapeutic investigations
3.5% (26/749) of all chimpanzee experiments.
61.5% (16/26) of these investigated the pharmacological
properties of various compounds.
Others included: the testing of surgical techniques or
prostheses, anaesthesiology and toxicology experiments.
50. Implications?
On the face of it, research on captive chimpanzees or
chimpanzee tissue appears to have contributed towards a
large array of biomedical disciplines.
However, …
51. Not all knowledge has significant value, nor is
worth the animal welfare-related, bioethical
and financial costs that may be incurred!
52. Statistically-significant subset of 95 chimpanzee studies:
49.5% (47/95) were not cited by any future papers.
Given that much research of lesser value is not published…
True conclusion:
The majority of chimpanzee experiments generate data
of questionable value, which makes little obvious
contribution toward the advancement of biomedical
knowledge.
53. Efficacy in combating human diseases
38.5% (34/95) were cited only by 116 subsequent papers that
clearly did not describe well developed methods for combating
human diseases.
Instead, papers focused primarily on:
non-human species ranging from bacteria to elephants,
including a large variety of primates;
human subjects in relation to a variety of biological disciplines
other than pathology;
examinations of the aetiological or other aspects of human
diseases.
54. Only 14.7% (14/95) of chimpanzee studies were cited by a total
of 27 papers that appeared to describe prophylactic, diagnostic or
therapeutic methods with sound potential for combating human
diseases.
Figure 4: Citations of
95 randomly selected
published chimpanzee studies
50 47
40 34
30
20 14
10
0
Not subsequently Cited by other Cited by medical
cited paper paper
55. Contributions of chimpanzee studies
63.0% (17/27) of these medical papers: wide-ranging reviews of
26-300 (median 104) references, to which the cited chimpanzee
study made a relatively small contribution.
Research methodologies contributing most:
in vitro studies
human clinical and epidemiological studies
molecular assays and methods
genomic studies
56. 12 cases: the cited chimpanzee studies appeared redundant, as
humans or human sera were studied concurrently, or because
they only served to confirm previous human-based
observations.
7 cases: the methods explored in the chimpanzee study were
not developed further, sometimes because later clinical trials
in humans failed to demonstrate safety or efficacy, contrary to
positive chimpanzee results.
57. 5 cases: the chimpanzee study examined a disease or method
peripheral to the prophylactic, diagnostic or therapeutic
method described.
Remainder: the chimpanzee study yielded results inconsistent
with other human or primate data, or merely illustrated
historical findings, or was only cited in order to discuss
concurrent human outcomes within the cited chimpanzee
study.
58. No chimpanzee study demonstrated an essential
contribution, or ― in a clear majority of cases ― a
significant contribution of any kind, towards papers
describing well-developed prophylactic, diagnostic or
therapeutic methods for combating human diseases!
59. 27 systematic reviews:
overall results
The authors concluded that the animal models were useful in
advancing human clinical outcomes or substantially consistent
with human outcomes in only 2 of 20 studies, and the
conclusion in 1 case was contentious.
7 reviews failed to clearly demonstrate utility in predicting
human toxicological outcomes such as carcinogenicity and
teratogenicity.
Consequently, animal data may not generally be assumed to
be substantially useful for these purposes.
60. Scientific limitations
of animal models
Differences between species and genders — with
subsequent effects on toxico- and pharmacokinetics (the
study of bodily distribution), or pharmacodynamics (the
study of mechanisms of action, and drug effects).
Loss of biological variability or predictivity resulting from
the use of in-bred strains, young animals, restriction to
single genders, and inadequate group sizes.
61. Lack of co-morbidities (concurrent illnesses) or
other human risk factors.
Physiological or immunological distortions
resulting from stressful environments and
procedures.
62. 1. Interspecies differences
Altered susceptibility to, causes and progression of diseases
Differing absorption, tissue distribution, metabolism, and
excretion of pharmaceutical agents and toxins
Differences in the toxicity and efficacy of pharmaceuticals
63. 2. False positive results of chronic
high dose rodent studies
Differences in rodent physiology when compared to humans,
for example, increased metabolic and decreased DNA excision
repair rates
Overwhelming of natural physiological defences such as
epithelial shedding, inducible enzymes, DNA and tissue repair
mechanisms, which effectively protect against many naturally
occurring toxins at environmentally relevant levels
64. Unnatural elevation of cell division rates during ad libitum
feeding studies
Variable, yet substantial, stresses caused by handling and
restraint, and frequently stressful routes of administration, and
subsequent effects on hormonal regulation, immune status and
disease predisposition
65. 3. Poor methodological quality of
animal experiments
At least 11 systematic reviews demonstrated the poor
methodological quality of many of the animal experiments
examined
None demonstrated good methodological quality of a majority
of experiments
Common deficiencies: lack of sample size calculations,
sufficient sample sizes, randomised treatment allocation,
blinded drug administration, blinded induction of ischaemia in
the case of stroke models, blinded outcome assessment and
conflict of interest statements.
66. Balls et al. (2004):
“…surveys of published papers as well as more anecdotal
information suggest that more than half of the published
papers in biomedical research have statistical mistakes, many
seem to use excessive numbers of animals, and a proportion
are poorly designed”.
67. Conclusions
The historical and contemporary paradigm that animal models
are generally fairly predictive of human outcomes provides the
basis for their continuing widespread use in toxicity testing
and biomedical research aimed at developing cures for human
diseases.
However, their use persists for historical and cultural reasons,
rather than because they have been demonstrated to be
scientifically valid. E.g., many regulatory officials “feel more
comfortable” with animal data (O’Connor 1997), and some
even believe animal tests are inherently valid, simply because
they are conducted in animals (Balls 2004).
68. However, most systematic reviews have demonstrated that
animal models are insufficiently predictive of human
outcomes to offer substantial benefit in advancing clinical
outcomes, or in deriving human toxicity assessments.
Consequently, animal data may not generally be assumed to
be of substantial use for these purposes.
The poor human clinical and toxicological utility of most
animal models for which data exists, in conjunction with their
generally substantial animal welfare and economic costs,
justify a ban on animal models lacking scientific data clearly
establishing their human predictivity or utility.
69. Animal experiments may sometimes be followed by a
concordant human outcome. However, they are a highly
inefficient means of advancing human health.
Additionally, they accord insufficient ethical weighting to the
interests of other species.
71. Replacement alternatives
Mechanisms to enhance sharing and assessment of
existing data, prior to conducting further studies.
72. Physicochemical evaluation and computerized modelling,
including the use of structure-activity relationships (predict
biological activity such as toxicity on the basis of structure),
and expert systems (seek to mimic the judgment of expert
toxicologists, by using known rules about factors affecting
toxicity, in combination with physicochemical or other
information about a specific compound).
These allow predictions about toxicity and related biological
outcomes, such as metabolic fate.
73. Minimally-sentient animals from lower phylogenetic
orders, or early developmental vertebral stages, as well as
microorganisms and higher plants.
74. A variety of tissue cultures, including immortalised cell
lines, embryonic and adult stem cells, and organotypic
cultures.
75. In vitro assays (tests) utilising bacterial, yeast, protozoal,
mammalian or human cell cultures exist for a wide range of
toxic and other endpoints. These may be static, or perfused,
and used individually, or combined within test batteries.
Human hepatocyte (liver cell) cultures and metabolic
activation systems offer potential assessment of metabolite
(product of metabolism, usually by the liver) activity — a
very important consideration when assessing toxicity.
76. cDNA microarrays (‘gene chips’) allow assessment of large
numbers of genes simultaneously. This may allow genetic
expression profiling (detection of up- or down-regulation of
genes, caused by exposure to test compounds). This can
increase the speed of toxin detection, well prior to more
invasive endpoints.
77. The safety profile and predictivity for diverse human patient
populations of clinical trials should be improved using
microdosing, biomarkers, staggered dosing, more
representative test populations, and longer exposure periods.
78. Surrogate human tissues and advanced imaging modalities.
Human epidemiological, psychological and sociological
studies.
Particularly when human tissues are used, non-animal models
may generate faster, cheaper results, more reliably predictive
for humans, yielding greater insights into human
biochemical processes.
79. Reduction alternatives
Improvements in experimental design and statistical
analysis; particularly, adequate sample sizes.
Minimising animal numbers without unacceptably
compromising statistical power, through decreasing data
variability:
Environmental enrichment, aimed at decreasing physiological,
psychological or behavioural variation resulting from barren
laboratory housing and stressful procedures.
Choosing, where possible, to measure variables with low inherent
variability.
Genetically homogeneous (isogenic or inbred) or specified
pathogen-free animal strains.
Screening raw data for obvious errors or outliers.
80. Meta-analysis (aggregation and statistical analysis
of suitable data from multiple experiments). For
some purposes, treatment and control groups can be
combined, permitting group numbers to be
minimised.
81. Refinement alternatives
Analgesics and anaesthetics. (Around 60% of UK
procedures are conducted without anaesthetics). While such
drugs undoubtedly alter normal physiology, claims that such
alterations are sufficiently important to hypotheses under
investigation, to warrant their exclusion, require careful
scrutiny.
82. Non-invasive imaging modalities.
Telemetric devices to obtain information remotely.
Faecal analysis (e.g. faecal cortisol monitoring).
Training animals (especially primates) to participate (e.g.
presenting arms for blood-sampling), rather than using
physical or chemical restraint.
Environmental enrichment.
Socialisation opportunities.
83. Increasing 3Rs compliance
Technology reproducibility and transfer: increased
methodology description, e.g., via publicly-accessible
databases, linked to scientific articles.
Redirection of public funds from animal models to
alternatives development/implementation.
84. Prerequisite for research funding, ethics committee approval,
and publication of results. Would require education and
cooperation of funding agencies, ethics committees and
journal editors about the limitations of animal models, and the
potential of alternatives.
National centre for the development of alternative methods.
Scientific recognition: awards, career options.
85. Likely benefits
Greater selection of test models that are truly predictive of
human outcomes
Increased safety of humans exposed to chemicals that have
passed toxicity tests
Increased efficiency during the development of human
pharmaceuticals and other therapeutic interventions
Decreased wastage of animal, personnel and financial
resources.
86. Conclusions
The scientific and logistical limitations incurred by the use of
animal models of humans within biomedical research and
toxicity testing are substantial, and increasingly recognized.
So is social concern about, and consequent regulatory
restriction of, laboratory animal use.
In defiance of these factors, such use remains enormous. The
use of GM animals, and the implementation of large-scale
chemical testing programs, are increasing laboratory animal
use internationally.
87. These trends clearly demonstrate the need for considerably
greater awareness of, and compliance with, the principles of
the 3Rs.
These principles are universally recognized as essential to
good laboratory animal practice, for animal welfare-related
and ethical reasons, and also, to increase the quality of the
research, and the robustness of subsequent results.
Published papers:
www.animalexperiments.info
www.aknight.info, ‘publications’