In my previous blogs, including Exercises in Simple Math Formulas, I have presented evidence that animal models are misleading and incapable of predicting human response to drugs and disease. I have also worked out the simple math equations for sensitivity and positive predictive value. Finally, I have provided references where more data amenable to calculating sensitivity and so forth can be found.
One thing I have failed to sufficiently explain apparently, is the fact that humans are the gold standard in the little 2X2 table. The gold standard is what the test or modality in question is compared to. Humans are also the gold standard when it comes to drug testing and disease research. That is, humans are the gold standard if one is studying diseases that occur in humans. Dogs are the gold standard for diseases of the canine persuasion and mice are the gold standard for murine diseases.
I would have thought the above would have gone without saying but, based on Dr Ringach’s statement, apparently I was wrong. Dr Ringach:
Actually, we do know that studying humans will result in information about humans. Some would call that—self-evident. Further we know that studying mice will yield information about mice. This has never been in doubt, at least in the minds of people whose livelihood and reputation are not dependent on research using animals.
The question is not, “Will research with human result in a medical advance?” but rather, “How large of a medical advance will come from properly conducted ethical research with humans or human tissue?” Likewise, the questions about animal-based research include:
- Can a nonhuman animal species predict human response to drugs and disease?
- Does research using sentient animals (as that is where society’s concerns lie) have a high probability of leading to new knowledge that will result in treatments or cures?
I have claimed to know the answer to both questions and have, in a very abbreviated form, answered both questions in my blogs.
The answer to Dr Ringach’s question about: “The probability that research based on animals can yield a medical advance” has been addressed thoroughly. We address it even more thoroughly in Animal Models in Light of Evolution. It is very low! This is not even debated in animal experimentation circles. The argument most animal experimenters use is that, despite the low probability that treatments will result from animal studies, they are the only intact system we can experiment on and that we can do things to animals that we cannot do to humans. (The first is fallacious in it’s own right but I will leave that for another blog. The second is true but raises the prediction question. I will blog on all this at another time.) My point is simply that the low probability is not an issue, even among Dr Ringach’s colleagues. In Animal Models in Light of Evolution, we flesh out this argument with examples and theory to put it all in context.
As I have said many times, the most powerful argument we have in support of our position does not, in my opinion, come from the empirical studies I have cited but rather from an examination of evolutionary biology and complex systems. I should here note that I think the empirical evidence is sufficient for deciding that animal models cannot predict human response and have a very low likelihood of leading to treatments. It is just that by putting it all in the context of evolution vis-à-vis regulatory genes and gene expression, add-ons as Ptashne and Gann (1) explain, evo devo, chromosomal duplication/alteration and duplication, gene deletions, insertion, inversions and duplication, copy number variant, SNPs, alternative splicing, the fact that the same gene can have different functions in different species, modifier genes and forth (2), one can develop an understanding of the theory. As I have said, understanding theories, or in the case of physics, the laws of physics, allow one to explain all of the empirical evidence and go beyond it.
Understanding the second law of thermodynamics allows one to skip reading all the patent applications for perpetual motion machines. Understanding evolutionary biology and complex systems allows one to conclude that animal models will never be able to predict drugs and disease response for humans. It also explains why studies like Contopoulos-Ioannidis et al (3), Lindl (4) and others (5, 6) show such a very low translation rate of animal studies into treatments for humans.
We are by no means alone in our assessment. For example, Hackam and Redelmeier of the University of Toronto Department of Medicine summarized their study that was published in JAMA October 11, 2006:
Only about a third of highly cited animal research translated at the level of human randomized trials. This rate of translation is lower than the recently estimated 44% replication rate for highly cited human studies. Limitations of this review include a focus on highly cited animal studies published in leading journals, which by their positive and highly visible nature may have been more likely to translate than less frequently cited research. In addition, this study had limited power to discern individual predictors of translation.
Nevertheless, we believe these findings have important implications. First, patients and physicians should remain cautious about extrapolating the findings of prominent animal research to the care of human disease. Second, major opportunities for improving study design and methodological quality are available for preclinical research. Finally, poor replication of even high-quality animal studies should be expected by those who conduct clinical research. (7) (Emphasis added.)
The numbers are there and they do not lie.
If you are interested in the subject of animals in science but the last few blogs have pretty much made your head swim, you might want to try FAQs About the Use of Animals in Science: A handbook for the scientifically perplexed.
I encourage interested, scientifically minded people to read the books referenced as #s 1 and 2 below.
1. M. Ptashne, A. Gann, Genes & Signals. (Cold Springs Harbor Laboratory Press, 2002).
2. M. W. Kirschner, J. C. Gerhart, The Plausibility of Life. (Yale University Press, 2006).
3. D. G. Contopoulos-Ioannidis, E. Ntzani, J. P. Ioannidis, Am J Med114, 477 (Apr 15, 2003).
4. T. Lindl, M. Voelkel, R. Kolar, ALTEX22, 143 (2005).
5. W. F. Crowley, Jr., Am J Med114, 503 (Apr 15, 2003).
6. J. Grant, L. Green, B. Mason, “From Bedside to Bench: Comroe and Dripps Revisited” (Health Economics Research Group. Brunel University, Uxbridge, Middlesex UB8 3PH, UK, 2003).
7. D. G. Hackam, D. A. Redelmeier, JAMA296, 1731 (Oct 11, 2006).