Of Mice, Men, and Medicine

Photo of someone holding the lab on a chip device next to a photo of two laboratory mice

Will a chip challenge the mouse?
Source: Wyss Institute and Bill Branson, NIH

The humble laboratory mouse has taught us a phenomenal amount about embryonic development, disease, and evolution. And, for decades, the pharmaceutical industry has relied on these critters to test the safety and efficacy of new drug candidates. If it works in mice, so we thought, it should work in humans. But when it comes to molecules designed to target a sepsis-like condition, 150 drugs that successfully treated this condition in mice later failed in human clinical trials—a heartbreaking loss of decades of research and billions of dollars. A new NIH-funded study [1] reveals why.

Sepsis is a life-threatening systemic infection. It can be caused by a variety of pathogens, including bacteria, viruses, and fungi. Serious consequences occur when tissues damaged by infection produce proteins sometimes called “alarmins” that send the immune system into overdrive. Traumatic injuries involving extreme blood loss or burns can set off the same dangerous response. To probe the molecular response to all of these triggers, the authors took periodic blood samples from 167 trauma (car crashes, falls) patients; from 244 patients with burns over at least 20% of their body; and from four healthy volunteers who had been injected with a low-dose bacterial toxin. Then they studied the activity of the genes in the white blood cells. Comparing the results, they found that of the 5,500 or so genes that responded to traumatic injury, 91% also played a role in burn response and recovery. And about 45% of these same genes were involved in recovery from the bacterial toxin exposure.

Mice, however, apparently use distinct sets of genes to tackle trauma, burns, and bacterial toxins—when the authors compared the activity of the human sepsis-trauma-burn genes with that of the equivalent mouse genes, there was very little overlap. No wonder drugs designed for the mice failed in humans: they were, in fact, treating different conditions!

But that doesn’t mean studying mice is useless. There’s still much the mouse might teach us. Mice, as the authors note, are more resilient to infection and mount a much more regulated immune response to pathogens than humans. While it takes relatively few bacteria in the bloodstream to make humans critically ill, it takes a million-fold more bacteria to sicken a mouse. Perhaps this is because mice nose around in some filthy places and can’t afford to overreact to every microbe? If we knew how these rodents limit the drama of their immune response, it might be useful for us humans.

But this study’s implications may well go beyond mice and sepsis. It suggests that we should not assume a mouse’s drug response will always accurately predict a human’s. It would be wise to monitor the activity of the genes and pathways of interest in humans and mice, to see whether a drug works the same way in the two species.

The new study provides more reason to develop better and more sophisticated models of human disease. More than 30% of all drugs successfully tested in animals later prove toxic in human trials. The NIH plans to commit $70 million over the next five years to develop “tissue chips”—miniature 3-D organs made with living human cells—to help predict drug safety and efficacy [2]. Though this is high-risk research, these chips may ultimately provide better models of human disease and biology than the use of animals.


[1] Genomic responses in mouse models poorly mimic human inflammatory diseases. Seok J, Warren HS, Cuenca AG, Mindrinos MN, Baker HV, Xu W, Richards DR, McDonald-Smith GP, Gao H, Hennessy L, Finnerty CC, López CM, Honari S, Moore EE, Minei JP, Cuschieri J, Bankey PE, Johnson JL, Sperry J, Nathens AB, Billiar TR, West MA, Jeschke MG, Klein MB, Gamelli RL, Gibran NS, Brownstein BH, Miller-Graziano C, Calvano SE, Mason PH, Cobb JP, Rahme LG, Lowry SF, Maier RV, Moldawer LL, Herndon DN, Davis RW, Xiao W, Tompkins RG; the Inflammation and Host Response to Injury, Large Scale Collaborative Research Program. Proc Natl Acad Sci U S A. 2013 Feb 11.

[2] Tissue Chip for Drug Screening

NIH support: the National Institute of General Medical Sciences; and the National Human Genome Research Institute

11 thoughts on “Of Mice, Men, and Medicine

  1. I wonder if we looked at it differently…rather than from human tissue to mice, is there research re: mouse (sepsis immune replication) to human tissue samples?

  2. Stated another way, more than 70% of all drugs successfully tested in animals succeed in human trials, making mouse research the most predictable and valuable avenue of success currently available to medical research.

    Come on Francis, don’t poison the minds of Study Section members against the thousands of valuable proposed studies.

    • The 30% is a typo. The failure rate for drugs successfully tested in animals is more than 90%, with at least five FDA and peer-reviewed references. You should know that.

      • Dr. Pippin,

        Thank you so much for bringing this discrepancy to the attention of Dr. Collins. During the production process, several key words were cut from the sentence in question. So, for purposes of accuracy and clarity, we have revised the sentence to read:

        “More than 30% of all drugs successfully tested in animals later prove toxic in human trials.”

        We apologize to readers of the NIH Director’s Blog for any confusion this may have caused.

      • Very interesting topic and discussions. Dr. Pippin above mentioned five FDA and peer-reviewed references showing a very high failure rate of drugs tested in mice and humans.Could you please give more details of these references? I’m very interested in this topic and found some references, but would like to be sure that they are the same. Very much appreciated.

  3. There is indeed hope that certain mice strains or genetically modified animals with humanized immune systems or 3D human tissue chips might prove to be more predictive for the human clinical situation. However, the problem remains to identify and develop these ‘new models’ and validate them for clinical development which will probably require pre-competitive consortia. Furthermore I think the fraction of failed drugs that work in mice but not in humans is likely more than 80-85% on average.
    The problem is hopelessly complex for CNS diseases; in fact for Alzheimer’s disease, over the last 10 years only 1 in 34 clinical development projects was succesfull (according to a recent study of the Pharmaceutical Research and Manufacturing Association), although all of these drugs showed a sufficient ‘preclinical signal’ in some animal model to invest large amounts of money. There is still a huge translational disconnect (see the recent report of the Institute of Medicine on ‘Animal models in CNS diseases’) and we might need looking at out-of-the-box approaches and technologies for solving these problems.

  4. These occasional blogs are highly informative and are written in a language that is easily understood and appreciated.

    Thank you so much.

  5. “Of Mice, Men, and Medicine,” reminds me of a comment I made in one of my E-mails ” Of Mice,and Men,” a John Steinbeck book and movie starring Candice Bergen, and Lon Chaney during depression days in California?

    Well, anyway you have to read the book or see the movie, but it goes to show the difference between mice and men all most a hundred years before genetic medicine realized that all creatures have a little different molecular genetic code …

  6. would be interesting to see if in the future these chips will eventually be fast enough to test someone almost immediately so that EMT workers or Paramedics could test someone at a scene of an accident to know what drugs they might not tolerate. interesting future for sure.

Comments are closed.