A recent study published in the Proceedings of the National Academy of Sciences calls into question the standard mouse model of sepsis, trauma, and infection. The research is an excellent example of how proper science investigates its own methods.

Mouse and other animal models are essential to biomedical research. The goal is to find a specific animal model of a human disease and then conduct preliminary research on the animal model in order to determine which research is promising enough to study in humans. There are also non-animal assays and “test tube” type research that is used to screen potential treatments, but scientists still prefer a good animal model.

It is also understood that animal models are imperfect – mice are not humans, after all. Animal research is therefore not a substitute for human research. I and other SBM authors have regularly criticized proponents of dubious treatments who make clinical claims based upon preliminary animal research. Until something is studied in humans, we cannot make any reliable claims about its safety and efficacy in people.

I would also point out that it is highly problematic to discuss the utility of animal models in general. Each animal model needs to be considered by itself, in terms of how predictive it is, and for what. For example, there are SOD1 mice who have a mutation that causes familial ALS (FALS). This mouse model has served as a screen for many potential ALS treatments. It turns out that the SOD1 mouse is an excellent animal model for SOD1 FALS in humans. This is not surprising considering that it is the same mutation. However, the SOD1 mouse model of ALS is much less predictive for sporadic ALS. Most of the drugs that look promising in the SOD1 model have not shown a significant clinical effect in humans with sporadic ALS (only one drug has actually made it through FDA approval: riluzole).

The question investigated by the authors of the current study is this – how good is the current mouse model of sepsis (infection in the blood), trauma, and other infections? The deeper question is – how similar is the mouse immune system to the human immune system.

The authors first questioned the mouse model after studying the pattern of gene activation in patients suffering from these three conditions, each of which cause inflammatory stress on the body. They found that the pattern of genetic response was similar across all these conditions, pointing to a common inflammatory response to stress. Their research was criticized, however, for not including mouse data to back it up. So they conducted follow up research looking at the pattern of gene activation to these stress conditions in mice.

What they found surprised them – mice had a different pattern of genetic activation in each of the three conditions, and all of them were different from the response found in humans. In other words, the mouse immune system responds differently to various kinds of stress than the human immune system.

The authors were quick to recognize the implications of their findings: the mouse model of inflammatory stress reactions is likely worthless or even misleading. According to their research, it should not be used as a preliminary indicator of human research into these conditions.

The data seems fairly robust and the conclusions valid. We still need time for the study to be digested by the scientific community, and perhaps for some follow up research to be conducted, but even with this one study it seems that researchers should think carefully about using mice to research potential treatments in these conditions.

Media reporting about this research was generally good, probably because the story is pretty juicy as it is, but as usually ramped up the drama a notch or two. It is common for researchers themselves to overestimate the importance and impact of their own research. Often they exaggerate prior ignorance and resistance to their findings in order to magnify the apparent effect of their new findings. The media will tend to focus on this aspect of a science new story, ramping up the drama even further.

The New York Time reporting on this study falls into this pattern, in my opinion. The author, Gina Kolata, makes it seem like the authors of the current study are the first to challenge the mouse model of inflammation, and were met with irrational resistance. She offers as evidence the fact that it took the authors a year to get past peer-review and that they were turned down by Nature and Science before being published by PNAS.

This is not unusual at all, however. Taking a year to get published is actually pretty good, and no one should cry about being turned down by two top-tier journals before being published in a third. This is not a sign of resistance, but more like standard procedure.

Further, there are previously published articles also expressing skepticism of the mouse model. This 2012 editorial, published in Nature Medicine and not by one of the authors of the current study, has the title: Rodent model of sepsis found shockingly lacking. This examination from 2008 also concludes that “Based on these criteria, the ideal model of sepsis does not exist.” This review concluded that there is utility to animal models, but they have significant weaknesses that need to be understood.

None of this is to imply that the current study is not an advance – the data on differences in genetic activation is very useful and does change the overall assessment of the utility of mouse models of inflammatory stress such as sepsis. I disagree with the impression given in reporting, however, that this notion comes out of the blue, is fundamentally different than prior attitudes, and was met with unwarranted resistance.


Animal models of disease are a vital technology to biomedical research, but they are a highly variable and problematic technology. This is well known to researchers, however, who seem to spend a sufficient amount of time questioning and studying the animal models themselves.

The current research appears to be a significant blow to the current mouse models of inflammatory stress, but also contains a wealth of information about which genes becomes active in humans and mice in response to such stress. This potentially can lead to other useful models or biomarkers, and also to new treatments in this very challenging area of medicine.


Posted by Steven Novella

Founder and currently Executive Editor of Science-Based Medicine Steven Novella, MD is an academic clinical neurologist at the Yale University School of Medicine. He is also the host and producer of the popular weekly science podcast, The Skeptics’ Guide to the Universe, and the author of the NeuroLogicaBlog, a daily blog that covers news and issues in neuroscience, but also general science, scientific skepticism, philosophy of science, critical thinking, and the intersection of science with the media and society. Dr. Novella also has produced two courses with The Great Courses, and published a book on critical thinking - also called The Skeptics Guide to the Universe.