It Turns Out Animal Models Really Aren't Predictive

Publish date:
Social count:

In the article A Long Way to the Bedside by Catherine Shaffer in Drug Discovery & Development, RBM CEO Ralph McDade was quoted as saying:

“Even a few years ago, people were still clinging to that blockbuster model,” says McDade. Nowadays, there’s recognition that linking a drug candidate to a strong biomarker gives it a better chance of success. “You have to identify the responders to your drug. It automatically cuts down the market size, but that’s the only way it’s going to get approved.”

The conceptual basis for personalized medicine is filtering down to the basic research community. In a press release from the Whitehead Institute for Biomedical Research:

So-called model organisms have long been at the core of biomedical research, allowing scientists to study the ins and outs of human disorders in non-human subjects.

In the ideal, such models accurately recapitulate a human disorder so that, for example, the Parkinson's disease observed in a rat model would be virtually indistinguishable from that in a human patient. The reality, of course, is that rats aren't human, and few models actually faithfully reflect the phenotype of the disease in question. Thus, in the strictest sense of the word, many "models" aren't truly models at all. To developmental biologist and Whitehead Institute Member Hazel Sive, this is no small matter.

"The term model is used very loosely," says Sive. "That was a problem to me: Everything's a model!"

Sive sees the need to adopt a new term to expand the language of biological research to encompass systems that, although not technically models, can still offer tremendous utility in studying the etiology of human disorders. In this setting, she proposes the use of the word "tool."

Sive, an advocate of using animals in research, outlines this idea in an editorial in Disease Models & Mechanisms. The press release continues:

Vivian Siegel, editor of Disease Models & Mechanisms, encouraged Sive to share it with the scientific community at large.

"She did exactly what I was hoping she would do." Siegel says of Sive's editorial. "We wanted to emphasize that you can find utility without the disease being recapitulated, as long as you know the limitations of the system." . . .

"If you adhere strictly to the term 'model', you can be misled by your own system," says Siegel.

When Niall Shanks and Hugh LaFollette published their work discussing the differences between using animals as causal analogical models (CAMs) and heuristic or hypothetical analogical models (HAMs) (LaFollette and Shanks 1995; LaFollette and Shanks 1996) the animal model community either ignored them or dismissed them. (Rowan 1997) (Shapiro 2004) Shanks, Jean Greek, and I have expanded the predictive versus tool concept (Shanks and Greek 2009; Shanks, Greek, and Greek 2009) and have been similarly dismissed. (See virtually any blog by Ringach, Simon Festing's review of Specious Science in the October 2002 RDS newsletter, or the ramblings of Orac.) But the above certainly seems to be consistent with what the three of us have been saying for decades: animals can be successfully used in numerous scientific endeavors, including as tools or heuristics, but not as predictive or causal analogical models.

In the final analysis the animal model community is going to be forced to accept the fact that animal models can in fact be used as tools but not to predict human response to drugs and disease because these are the facts of the material universe. They will suffer a financial loss, in the form of loss of funding, and their collective ego will take a hit in the form of loss of respect but those are the consequences of making false claims all these years. For example, Freeman and St Johnston in the same journal Disease Models & Mechanisms 2008:

Many scientists who work on model organisms, including both of us, have been known to contrive a connection to human disease to boost a grant or paper. It’s fair: after all, the parallels are genuine, but the connection is often rather indirect. DMM is about something quite different. This new journal is aimed at people who set out with an explicit goal to investigate human disease using model organisms. (Freeman and St Johnston 2008) (Emphasis added.)

Society will need to rethink whether it wants to allow various experiments on animals in light of the fact that the researchers are doing basic, not applied research, and hence the results will not make the drug supply safer and curing cancer in mice will not be synonymous with curing cancer in humans. (Greek and Greek 2010) This is how animal-based research has historically been sold to society and, as I have often said, such claims are simply not true.

But in the end all this will happen because science is the best way humans have of understanding the material universe and the science on this is very sound. The vested interest groups will scream and yell, but in the end science will prevail.


Freeman, Matthew, and Daniel St Johnston. 2008. Wherefore DMM? Disease Models & Mechanisms 1 (1):6-7.

Greek, R., and J. Greek. 2010. Is the use of sentient animals in basic research justifiable? Philos Ethics Humanit Med 5:14.

LaFollette, H, and N Shanks. 1995. Two Models of Models in Biomedical Research. Philosophical Quarterly 141-60.

LaFollette, Hugh, and Niall Shanks. 1996. Brute Science: Dilemmas of animal experimentation. London and New York: Routledge.

Rowan, Andrew. 1997. Book Review. Brute Science. Animal Welfare 6:378-81.

Shanks, N, and R Greek. 2009. Animal Models in Light of Evolution. Boca Raton: Brown Walker.

Shanks, N., R. Greek, and J. Greek. 2009. Are animal models predictive for humans? Philos Ethics Humanit Med 4 (1):2.

Shapiro, KJ. 2004. Animal Model Research. The Apples and Oranges Quandry. ATLA 32 (Suupl 1):405-09.


Popular Video