Wednesday, 13 February 2013

Failure to Translate. Another Nail in the Coffin

A cornerstone of modern biomedical research is the use of mouse models to explore basic pathophysiological mechanisms, evaluate new therapeutic approaches, and make go or no-go decisions to carry new drug candidates forward into clinical trials. Systematic studies evaluating how well mouse models mimic human inflammatory diseases are non-existent. Here, we show that, although acute inflammatory stresses from different aetiologies result in highly similar genomic responses in humans, the responses in corresponding mouse models correlate poorly with the human conditions and also, one another. Among genes changed significantly in humans, the mouse orthologs are close to random in matching their human counterparts (e.g., R2 between 0.0 and 0.1). In addition to improvements in the current animal model systems, our study supports higher priority for translational medical research to focus on the more complex human conditions rather than relying on mouse models to study human inflammatory diseases.
What does this study have to do with MS?, well nothing. However it has an important point. For many years the mantra has been we need a mouse model, especially a gene knockout/transgenic model to study human disease. 

In trials on septic shock there have apparently been 150 trials and no translation for human benefit, for EAE we have a few thousand drugs and few have translated into treatments of autoimmunity. This is a common theme in every subject that there is no translation from animals to man

Are all the models just pants?, or are there other issues like humans are not that great in designing and performing mouse studies? or are clinicans just rubbish at designing effective trials? Whilst, I am pretty sure the last two statements are a large part of the problem, because in humans you usually get the treatment hours or days after the problem starts, when it is too late and in animal studies the drug goes in before you induce the problem. However, in this study it points to problems with the preverbial beasty.

They look at the white blood response in people with severe burns, and trauma  hours, days and a year after the event. They then do the same to the poor animals (25% total body burn-come on guys... really why such reflect a human group then select humans with fewer burns) and look at the response over a few hours to a week, without mention of analgesic or whether they were actually kept under anaesthetic for the whole week after the period of scald. They even get human volunteers to get a bacterial toxin and induce that in mice also. 

They looked at what gene pathways are activated and it turns out that human inflammation, with what ever the cause, has similar influences in regulation of different gene pathways. Likewise, gene pathways are similarly induced during mouse inflammation, irrespective of the cause, but gulp the human response looks very different from the mouse response and the recovery times differ and are much quicker in mouse compared to humans.
            The paper is open access have a read for yourself

Therefore, the inference is that mouse models are pants and we should just focus on complex human conditions and not do animal studies, becuase they do not reflect human responses. They suggest that new approaches should be taken and first genomic approaches should be undertaken in human disease to define human disease and this could be used to define the animal model. Synthetic human models could be developed with disease related cell types.

I have no problem with this approach and applaud it. However some form of model is needed because this is part of the drug developmental process that helps pharma commit large resourses for treatment trials.

Within Immunology in MS, a powerful lobby has dictated that the animal model has shaped the way that clinicians approach MS....It is the tail wagging the dog. For progressive MS, it is important that we learn for this experience and get the dog to wag the tail. Indeed it was thought even before the IPMSC meeting last week, which discussed this matter, that examination of human pathology should be used as a cornerstone to define help inform and define the models.

Is the genomic approach the way to go? It appears the case that this divergent set of pathways can result in a similar cytokine phenotype. If the problem is that a man from London goes to Manchester, does it matter if they travel via Birmingham in a car or go via Leeds in a train, the man gets to Manchester in the end. If you stop the man getting to there then you deal with the problem.

The authors of this post argue that the solutions to all problems is a genomic approach. However, the genomic approach has not enlightened us a whole lot about MS susceptibility so far. It tells us that the main genetic aspect is the Major histocompatibility Complex, which we knew over ten years before the new genomics era and that there are over 50 genetic components that have some influence in MS susceptibility. PPMS and RRMS can apparently develop in identical twins, so genetics alone may not be the whole issue on how progression develops.

It would have been relevant to know how different the baseline result is because things are measured as fold increases, if an animal already has high levels of message X there may be no need to increase it. I suspect that they are different to start with, so same conclusion without the suffering that message levels in mouse and humans are different.

The mousers response to this work is that the results in mice essentially come from one strain of mouse. What happens in other strains of mice? Whilst this cop-out is a possibility, it is probably very unlikely to make much of a difference, although the strain they used is one of the worse strains in terms of suceptibility to use for EAE research. Rodents are highly resistant to most types of induced inflammation compared with humans. They live in holes in the ground and spend most of their lives in the wild rumaging around rubbish (Garbage in American) so they have evolved to deal with dirty, unhealthy situations.

The dose of bactrial toxin used in most published in vivo studies in mice is 1–25 mg/kg, which approximates the dose that causes death in half of the mice. This dose is 1000–10,000 times the dose required to induce severe disease with shock in humans and ∼1,000,000 times the dose used in carefully controlled study environments in which human volunteers are challenged with highly characterized preparations of 2–4 ng/kg bacterial toxin to elicit fever and cytokines, as used in this study. A direct comparison between the species for live bacteria is obviously not possible. However, in most mouse models, it is necessary to administer high doses of live bacteria to induce illness (usually 10 billion) and whether this difference in sensitivity is the reason that severe infection in mice has little resemblance to sepsis syndrome in humans remains to be established by seems plausible. Guinea pigs are much more sensitive to the effects of bacterial toxin. I wonder if they are more human like?

The question is now whether this difference is true for all disease processesAre the processes involved in progression likewise very divergent between rodents and humans? Who knows until they are tested. 

However, this is yet one more nail being banged into the coffin of animal experiments as some will just remember the take home message that "Mice lead scientists astray"  

I wonder if many people remember this one? "S**t-in, *hi*-out" so it is important that we do experiments right.


  1. Interesting, I saw this online and thought I would come here to see what you had to say about this. I did notice that mice are much more resilient to bacteria, viruses and fungi and that makes me wonder how the three of them have been determined in people with MS. I have often wondered how it has been determined that Lyme, CPN, fungi, viruses have been ruled out as the cause for MS or if there are current studies? Has there been published peer reviewed studies on these and how these things have been ruled out or not? Thanks

  2. Viruses have certainly not been ruled out, if they had Prof Gs would not be doing the Charcot project where viruses are at the forefront of their thinking...keep an open mind, but some pathogens do not fit with the biology or the reality.

    There have been peer review stuff on most things and their are plenty of armchair scientists out there making their peer-reviewed ideas public

    1. Looked at the on-line stuff and there seems to be more of a gripe that "Nature" and "Science" first rejected the paper. Boo hoo. What relevance is that to the content...nothing it is sour grapes....they have to deal with it like most of us deal with rejection from those journals.

      You pick yourself up and move on, if I griped about the number of journals that our tolerance paper,with a immunological route to a cure, went to that would be half a page.

      The world was not ready for it then and I suspect with the mousers are not ready for this current paper and will carry-on regardless.. However, the funders will be taking note and so you may find it that bit harder to get funding for that immunological study. However, if truth be told, this approach has been difficult for a while now as it has not been where priority for MS has been at least in the UK.

  3. Trouble is that when papers like this are published there can be a rapid jump to the Baby/Bathwater syndrome.

  4. Yep just like the refining EAE post. The intention was to do good to produce practical solutions to reduce suffering but with the "idea to reduce the severity classification from substantial" the intention would have to be replacement as the severity is inherent in the condition. They served to indicate the disease may be worse than it probably is and so it will be harder to convince ethical committees to support it, without more hurdles. Another nail in the coffin of UK animal research. It must have been some ethical committee to support 25% total body area scald,

    The way to reduce the severity of EAE is to change the model or change the endpoints such that some experiments can be terminated early such as through Kaplan Meirer analysis to time to disease onset. Therefore animals would not need to become paralysed. Would the community have adoped this, well if they were asked they could of had a debate.

  5. As a "mouser" I must comment that the issue of mouse strain is critical and should have been considered much more closely by the authors of the PNAS paper. As a developmental biologist (I basically study how an egg becomes a baby, which is remarkably similar across species), I can attest that different inbred strains can behave very differently...each inbred mouse is almost a clone of its littermates, so it would be like looking at a single patient and drawing sweeping conclusions from the data. These authors are all very seasoned clinicians and scientists and had to have known that the mouse strain they selected was a poor choice. Beyond that, however, this paper had numerous other problems that escaped notice of the reviewers, editors, and many readers, apparently. For example, they looked at total blood leukocytes without identifying specific cell comparing bowls of fruit without knowing which fruits are in there or how many of each (ie, apples to oranges)! Their statistical analysis is flawed: whoever heard of a negative R-squared value?! This is a glaring and careless error that should have been caught. They also relied solely on gene chip-based microarrays and did not validate their data using any other method, which is also careless...what are the proteins doing? Plenty of "gene" regulation happens after the RNA is made, and this was ignored entirely.
    Now, ask yourself why these seasoned scientific veterans would want to publish a paper saying that animal models are useless. They said it themselves: They want more emphasis (and hence funding for) clinical studies!! Why waste time with animals when you can jump right to humans!? Except maybe that it costs a ton more, and can potentially put lives at risk. I suspect that there will be a lot of fallout from this study, but encourage everyone to wait for the dust to settle a while before adding any more nails to that coffin!

    1. I agree with many of the points you are making, and totally agree about the strain thing.

      As you say, doing experiments on an inbred animal is like doing the same experiment on the same person over and over again and thinking that it tells you how the total population will behave.

      Indeed it is unfortunate that gene knockouts are made on the C57BL/6 background which is about the worse genetic background for experimental MS studies. It means I believe that loads of published papers are probably meaningless in terms of any biological relevance to treating MS. (I know I will need to back this up with data but they are too easy to treat. This will be submitted in due course once other papers are dealt with).

      I take your point that there may be flaws in the study, but this point will already be lost as there is a tide building against animal experiments. This is the clinical view point and it is gaining increasing momentum. Fewer and fewer animal studies are getting funded.
      at least in the UK.

      Unfortunately the self-righteous clinicians do not look at themselves with such a critical eye. They mess up all the hard work of the "mousers" but rather than say the clinical work is flawed the animals studies are blamed.

      There are many "teflon men" in science...where s**t don't stick. This is another rant on the back burner, which I will address this also once other stuff has been dealt with.

      However,I also have to say, the head in the sand approach of many experimentalists will not help their cause. There is a load of old pants published week on week, much of if in the so called quatity rags. They want popularist nonsense and that is why about 70% of it can never be repeated.

      I hope you are right and we wait before more hammer work goes on.

  6. Great comments from anon 5 April 16.00. I too suspect a not too hidden agenda here. Shame you didn't get the paper to referee!
    Time for another ranting paper from the mouse docs!

  7. My agenda is not so hidden...I am a researcher in "translational" medicine, with a focus that overlaps that of the paper in question. I have a PhD, not an MD, so I cannot run clinical studies. Also, for academic labs in the US (where I am) it's nearly cost-prohibitive to try a clinical study on your own. The FDA rules and requirements are set up with pharmaceutical companies in mind, so it costs a fortune for any academic research lab to do clinical studies. Apart from the focus on need for funding, it's simply unethical (in my opinion) to attempt clinical studies of a therapy that doesn't have some basis established from prior preclinical or animal experiments. I know this isn't always possible, but that doesn't mean we don't need animals...we just need better models. For some diseases or conditions, there is simply not a large enough patient population to perform meaningful clinical studies (ie, with sufficient statistical power and proper controls)...that doesn't mean the problem isn't worth addressing.
    Also, I've published several microarray studies and am very picky about the analysis, so I was pretty shocked by the broad and bold statements by the PNAS paper authors. Not a big help to any translational researcher, I'm afraid.
    Thanks for your comments.

    1. Unfortunately P-NAS and (load of old) b*ll*cks can go together :-)

      PNAS is the Proceedings of the National Academy of Science of the USA.
      A prestigious not say more.

      Having spent all day trying to work out how to submit something there,
      the page length prediction has been killing me

    2. "I have a PhD, not an MD, so I cannot run clinical studies".

      Problem is many MDs cannot run proper trials either.

    3. "but that doesn't mean we don't need animals...we just need better models." Never was a truer word spoken. We bang our heads against the same brick wall (sometimes we manage to headbutt a brick lose). To simply dismiss animal models as not worthwhile is to my mind lazy and arrogant.

      Q. What's the difference between some MDs and a computer?
      A. It's a lot easier to punch information into a computer! ;-)


Please note that all comments are moderated and any personal or marketing-related submissions will not be shown.