AFMA and the October 2008 issue of Nature Medicine

The August 2008 issue of Nature Medicine (NM) featured an editorial about the use of nonhuman primates in research. NM is a very prestigious journal. According to their website: “The 2007 impact factor for Nature Medicine is 26.382, according to the Thomson Reuters Journal Citation Reports. This places Nature Medicine as the top primary research journal in Medicine (Research and Experimental).”

Shanks and Greek wrote a letter regarding the editorial that was published (the correspondence can be viewed at http://www.nature.com/nm/journal/v14/n10/full/nm1008-1012a.html if you have a subscription or are willing to pay to download). Two other letters were published regarding the use of nonhuman primates and the editors then responded to all three of the letters. AFMA would now like to address that response.

Before we examine the issues raised, AFMA would like to thank NM for printing the letter and for addressing the issue at all. Most journals are not as courageous and NM stands out as the leader they are for addressing the issue and allowing responses.

Shanks and Greek stated, with supporting references, that nonhuman primates, while viable for obtaining knowledge are not predictive models for humans. NM then stated, in part, the following which is very typical for the industry.

1. Animal models while not being predictive are useful.
2. Even if the model does not predict human response it can still be used, as it is the best research method we have. In vitro and in silico are not predictive.
3. Even if the model does not predict human response it can still be used, as the more we understand about the model the more we can make the model mimic humans.
4. They cite the “usefulness of experimental autoimmune encephalomyelitis as a model of multiple sclerosis . . . because successful therapies have emerged from their use.”
5. Animal models should not be abandoned because it is very difficult to develop new drugs.
6. It took 50 years to develop a polio vaccine so we should not expect great breakthroughs like an HIV vaccine to occur quickly.

We will address the above in order.

1. Animal models while not being predictive are useful.

In Shanks and Greek’s letter they acknowledged that animals could be used in science for many purposes. Their only point was that animals such as nonhuman primates are not predictive. The editors at NM are changing the subject when they justify using animals because they are useful. The issue here was prediction not usefulness. Further what do they mean by useful? Useful how? Useful for whom? Any research method can be useful in the most general sense of the word, but that does not mean it should be funded. To attempt to justify animal models because they are useful is to cloud the real issue of predictability which was the point Shanks and Greeks were making.

      2.  Even if the model does not predict human response it can still be used, as it is the best research method we have. In vitro and in silico are not predictive.

It is true that in silico and in vitro are not predictive for drug toxicity and other areas however neither are animal models and in vitro and in silico using human tissue does give us a true picture of what is happening in humans in disease research even its uses are limited. Using in vitro techniques to study HIV led to the 2008 Nobel Prize in Medicine or Physiology. Such research methods are underfunded compared to animal models. However, the relative merits of various research methods are immaterial to the issue of predictability.

If the purpose of the model is to predict human response and the model does not accomplish this, then the model has not fulfilled its purpose. Whether another technique can fulfill the purpose is irrelevant as the question remains, “Does the animal model predict human response?” Either it does or it does not. Astrology cannot predict the future therefore it is useless for that purpose. Just because no other modality can predict the future does not justify using astrology to see into the future.

The burden of proof for using animals as predictive models is on the animal modelers themselves. Animal models are sold to society on the basis of prediction. If they are not predictive then the scientific community should stop telling society that animal models will make their drugs safer (by predicting human response) and help us cure diseases (again by predicting how the disease affects humans).

     3.   Even if the model does not predict human response it can still be used, as the more we understand about the model the more we can make the model mimic humans.

This assumes we can make the model more accurately resemble humans. Genetically modified animals have shown this to be false. Davis Horrobin [D. F. Horrobin, Nat Rev Drug Discov 2, 151 (Feb, 2003)] wrote in 2003:

Does the use of animal models of disease take us any closer to understanding human disease? With rare exceptions, the answer to this question is likely to be negative. The reasoning is simple. An animal model of disease can be said to be congruent with the human disease only when three conditions have been met: we fully understand the animal model, we fully understand the human disease and we have examined the two cases and found them to be substantially congruent in all important respects...

Will genetically modified mice lead to better understanding of human disease? The only appropriate answer at this stage is perhaps or perhaps not. But the omens are not good and the confidence of so many in the Castalian establishment seems to me to be entirely misplaced. Why am I so sceptical?

First, most human disease is highly unlikely to be due to a single abnormal gene. It may well be that the consequences of catastrophic failure of a single gene can be partly understood with the assistance of appropriate genetically modified mouse models. But such diseases are for the most part rare and tend, in any case, to be reasonably well understood from direct human studies.

Second, consistent phenotypes are rarely obtained by modification of the same gene even in mice. The disruption of a gene in one strain of mice may be lethal, whereas disruption of exactly the same gene in another strain of mice may have no detectable phenotypic effect. If this is true of the impact on one gene of the rest of the mouse genome, how much more is it likely to be true of the impact of the rest of the genes in the human genome?

Third, the great majority of human diseases that affect large numbers of the population are likely to be the result of the interaction of several different genes. If one mouse gene is so difficult to understand in a mouse context, and if the genome of a different inbred strain of mouse has so much impact on the consequences of that single gene's expression, how unlikely is it that genetically modified mice are going to provide insights into complex gene interactions in the non-interbred human species? At the least one must conclude that most predictions of near term human benefit are not only overblown but are actually fraudulent.

     4.   They cite the “usefulness of experimental autoimmune encephalomyelitis as a model of multiple sclerosis... because successful therapies have emerged from their use.”

This is problematic for several reasons. 1) Just because a treatment followed temporally the use of an animal model does not mean the use of the animal model caused the treatment to be discovered. 2) If the animal model was successfully used was it necessary? The position of the animal model industry is that such models are necessary for science to discover cures and treatments. 3) A review of the literature casts doubt on whether the experimental autoimmune encephalomyelitis model of multiple sclerosis has been useful at all in explaining human multiple sclerosis. The burden of proof for claiming it has been useful in this fashion is on the person making the claim.

   5.   Animal models should not be abandoned because it is very difficult to develop new drugs.

It is difficult to develop new drugs, that is why society and shareholders should demand tax, charity, and company dollars be used only on scientifically viable research methods. Just because something is hard does mean we abandon science and the scientific method in search of answers.

  6.   It took 50 years to develop a polio vaccine so we should not expect great breakthroughs like an HIV vaccine to occur quickly.

Science is more advanced today than in the first half of the 20th century when the search for the polio vaccine was waged. It is not unreasonable to expect faster progress in light of more knowledge. Further, one reason it took so long to develop the polio vaccine was the use of animal models that reacted very differently than humans. Those who do not learn from history are doomed to repeat it.

These themes are explored more fully in Animal Models in Light of Evolution by Shanks and Greek, due out in 2009.

Again, AFMA thanks NM for addressing these issues and gratefully acknowledges their leadership in the field.