Of Mice & Men: My Rodent Study Disclaimer

The purpose of this post is to have a sort-of "disclaimer" to link to (if I can remember) whenever I discuss a rodent study here at the Asylum ... A small manifesto if you will, on my feelings on the utility and limitations of rodent studies in relation to human beings.   Also, I look at this post as a way to remind older readers and/or inform newer readers as to my background.  I spent roughly five years in the company of countless rats and mice, working for Big Pharma, and when I speak of rats, I'm not talking about my colleagues ;-)  My job was half analytical chemistry, half metabolism.  I worked on mostly discovery level drug candidates (a level before development) characterizing the absorption, metabolism and routes of excretion of these drugs, mostly antidepressants.  As such I've chopped the heads off of more rats than I care to recollect, opened up countless abdomens of mice to get maximum blood samples from the vena cava, removed/homogenized and analyzed countless livers, brains and intestines, observed behavior of these animals in their cages when dosed with the drugs I was giving them, etc.etc.etc.   As such, I do take somewhat personal affront to those who issue blanket dismissals of any animal studies.  As far as I'm concerned you're all a bunch of hypocrits unless you have never taken anything more than an aspirin and/or never intend to and plan to die a "natural death" feeling full pain and discomfort from whatever might befall you on this journey we call life.  I'm not going to turn this into a defense of Big Pharma, I have no affiliation, and since the days when I worked there (late 80's) I've felt our country is vastly over-medicated (and it has only gotten so so much worse since).  But I mention this because I believe my perspective is a bit unique in this regard.

OBVIOUSLY, I am not a mouse or a rat, and neither are you.  But, in many metabolic pathways, we are shockingly similar.  There are all of these highfalutin discussions about this blogosphere we inhabit, where enzyme acronyms are thrown about as frequently as HFCS.  They would not be going on were it not for the bazillion rodents who gave their lives so that these enzymatic pathways and receptors of various hormones, not to mention the hormones themselves, could be discovered and characterized.   I can't for the life of me fathom the logical contortions going on in the mind of someone who declares "I am not a mouse" out of one side of their mouths, while touting the imperative of a ketogenic diet based on studies in 1mm long earthworms who eat E.coli!  At least a mouse has a brain and a face for crying out loud, and at least it eats, or can eat, something remotely equivalent to human foods.  By the same token, unless at some point the findings in rodent models can be replicated in humans, they will forever bear the "scarlet letter" of being "only in rodents" ... and rightly so.  As such, we need to be careful what the findings tell us.  Oh ... and yes, there are differences between rats and mice, but for the purposes of this post, I'm probably going to use mostly mice.


Similarities:  Rodents are omnivores and can eat "human" diets that they chew and swallow.  They share many human characteristics:  brain, eyes, ears, similar digestive system (differing in the distal intestine), pancreas, liver, etc.etc.  They are warm blooded land animals that don't hibernate seasonally or developmentally.  They share many metabolic pathways, from mitochondrial metabolism, glycolysis, beta-oxidation of FA's, gluconeogenesis, de novo lipogenesis, glyceroneogenesis, etc. with humans.  They also share various hormones and signalling peptides, immune systems and functions, etc.

Ethics:  The bottom line for humans, we do not experiment on other human beings.  I do not know that Keys' famous Minnesota starvation experiment would be allowed today -- especially given some of the tragic outcomes for some of the participants.  So we can do many things with rodents that we could simply not even begin to think of trying in humans, let alone actually doing so.

Control:  Since we're mostly talking metabolic studies for this discussion, rodents offer us the opportunity to control intake, light cycles, temperature, whatever, essentially from birth-to-death.  A rodent can't cheat on its diet.  A rodent can be force-fed by gavage.

Lifespan:  Rodents have a shorter lifespan, mice especially, in which to observe the effects of aging on metabolic processes and/or to assess longevity.  Also, where a human study might involve months to see changes, this is only a matter of weeks in mice depending on the stage of life.  

They Are Not Human:  When it comes to obesity, body image, comfort foods, etc., none of this stuff applies to a rodent.  Looking at eating behavior, these animals eat when hungry, stop when full for the most part.  So we do get to see where metabolic balance is disturbed (e.g. the ob/ob mouse) without artificial constraints.  You can restrict calories and we have no idea if the mouse is hungry or that it notices it's belt can be tightened another notch.  When you weigh it, the mouse doesn't care what the number says and isn't likely to be depressed all day if he's gained a gram.

Genetic Modifications:  The sheer volume of information garnered from the creation of various "knockout" mice, etc., is staggering ... and that is an understatement.  We could not do this with humans.  Hopefully ever. The various KO mice allow us to isolate the action of a single gene, protein, receptor, etc. in an otherwise normal living being.  This it like in vitro on steroids!


They Are Not Human:  What may work with mice in a controlled environment may not work "in real life" for humans.  The various free living diet comparison studies pretty much tell us that.  Our reality includes many reasons besides pure physiological hunger driving intake, we care about or at least notice our waistlines, etc.  I don't imagine something as innocent as one mouse squealing to another that they look like they're having a bad fur day ...  There are also numerous enzymes, hormones, etc. that differ in form, function and significance that number too many to even quantify.

Body Size & Metabolic Rate:  I've shown the graphic at right, or something similar, here before.  Small mammals expend an awful lot more energy per unit mass than larger ones (like us humans) do.  Therefore a larger proportion of their metabolic rate is devoted to thermogenesis or generation of body heat as compared to locomotion, etc.  The changes in metabolic rate for thermogenesis and energy conservation are outside the realm of conscious intervention -- IOW we cannot control these factors in the "CO" side of the calorie balance equation.  In mice these adaptations are likely far greater than would be seen in humans.

Fat Type & Distribution:  Related to body size/thermogenesis needs, mice have far more brown fat (BAT) containing those uncoupling proteins that "waste" calories, UCP1.  Although adult humans have more BAT than previously thought, it is still not much in comparison.  Mice also have far more visceral fat (VAT) -- within the abdomen -- than subcutaneous fat (SAT) than humans.  The major SAT depot in mice is intrascapular -- between the shoulder blades on the back of their "neck".  Humans, women in particular, have significant SAT stores in limbs and elsewhere.  The nature of the metabolic activities of SAT vs. VAT is quite different and this must be considered when attempting to extrapolate mouse data to humans.

Body Shape & Mass Distribution:  Us biped humans have long arms and legs relative to the "body" and have considerable muscle mass, bone, etc. in our limbs.  Mice, on the other hand, are quadrupeds with short limbs and not a whole lot of lean mass involved.  Humans will expend more energy in activity because of these "mechanical" differences.

Image URL:  http://media.wiley.com/CurrentProtocols/PH/ph0558/ph0558-fig-0005-1-full.gif
Digestive Differences:  Mice have considerable differences in the distal small intestine/proximal large intestine area.  In humans this is not a well-defined feature, but rather called the ileocecum or cecum in the  beginning of the large intestine that may or may not include the distal small intestine depending on the diagram you're looking at.  Mice have a defined cecum as they are hind-gut fermenters.  The picture above shows an inflamed cecum but it should still give an idea of the size of this "organ" compared to the intestine.  As such, mice "digest" far more indigestible matter via gut biota fermentation than humans.  Mice may also engage in coprophagy -- eating feces -- and their metabolisms are much more subjected to gut flora population and type than humans apparently are.

The Role of De Novo Lipogenesis:   Perhaps it is due to the larger-scale production of short chain fatty acids in mice that DNL is a far more quantitatively substantial metabolic pathway in the mouse than humans.  This is also perhaps why mice fatten so easily on high fat diets, often despite eating similar calories, than high carb diets, because it is energy intensive to convert carb to fat for storage, whereas storage of LCFA is rather efficient.

Lifestages and Growth:  While humans can continue to grow past puberty, most tend to reach maximal height by their mid-to-late teens by the time we are reproductively adults.  Mice become reproductively adult long before their growth rate slows, and it is said that they continue to grow throughout their lives.  I would qualify that statement to address that at some point in longevity studies, the size/mass levels off and does indeed decline.  However in many studies the mice we are dealing with are in the 8-10 wk old range, reproductively adult, but still growing considerably, often for the duration of studies that may last several months.  This does not happen in humans.

Gen Mod/Knockouts are Not "Normal":  Let's take the ob/ob mouse that is totally leptin deficient.  If this occurs in a mouse it becomes obese, we've bred millions of these critters for obesity studies.  The human equivalent of the ob/ob mouse apparently does not exist in any considerable number in the population.  Therefore, although some obese humans may be analogous to the ob/ob mouse, this cannot provide many clues into what has caused the obesity epidemic.  Knockouts that alter one isolated function often produce widespread changes in the whole organism.  These are rarely analogous to actual conditions so must be viewed with caution.  I am planning another manifesto of sorts on just these sorts of issues and will edit in a link to that when it is complete.


One reason for this post was to put this general disclaimer on all citations of mouse (or rodent) studies:
This study was conducted in mice (or rats), and therefore the results should be viewed and interpreted appropriately as to whether they have any relevance to human metabolism.
Each case is different, and should be considered in context!  That said, I call for a ceasefire in the war of selective citations of animal studies.  The same people who decry the idiocy of those of us who don't buy into TWICHOO, and complain mightily when we discuss any animal study, need to be consistent within their own ranks.  Firstly, they tend to misrepresent how folks like myself and others view the utility of these studies to begin with, and secondly they often attribute beliefs to same that are not what have been expressed in a blog post, etc.  This wouldn't be quite so offensive if they didn't tout, link to and sarcastically celebrate when their favorite blogger has posted about ... another mouse study.  

You can't have it both ways.  And if folks want to be taken seriously on this issue they need to be consistent.  There are instances where mouse models can tell us useful information -- and where that information has been demonstrated to hold up when studied in humans.  There are also studies that are so far from the reality of what we're looking at that they offer little or no useful information, particularly "actionable" information.  

The next time Taubes stands up and shows the slide similar to the fat rat at right, trying to blame insulin for the obesity, let's hear the boos please.  And a little scrutiny when people stretch the envelop of what a study actually demonstrates would be nice.  I expect it of readers here too.

This post is by no means all-inclusive.  I'm sure I left some things out.  Please feel free, as always, to contribute in the comments on any issue you think I forgot, got wrong, or need to expand on a bit.  


LeonRover said…
I look forward to a study on AFLD and the Napa Valley with the heading

"The Grapes of Wrath"

Sanjeev said…

I lean more toward never wanting to use rat/mouse studies, but a well thought-out, well-laid-out roadmap of what information is more likely applicable, and which human mechanisms/pathways/comparisons are less likely to be illuminated (if at all) is a must if one is to use other-species studies.
LeonRover said…
More seriously: I emphatically agree with "This it like in vitro on steroids!".

Thus the existence of a mechanism may be shown in some living organism.

However, until that mechanism has been shown to be induced in humans, it is merely interesting.

Some researchers have used results from nematodes to embark on life-long caloric restriction.

Not this boy, my Dad lived to 91, and never calorically restricted himself. I judge that his genetics and epigenetics have been expressed in me.

"posterity be praised"
I'm not only hesitant to apply mouse studies to humans. I'm sure studies of Chinese humans don't necessarily apply to Caucasians. And studies of African-American females may not apply to white American men.

LeonRover said…
And Steve, what about when results from male studies are applied to women (and vice versa) even within racially similar groups.

For example there is no believable explanation, in my view, of the differences in age related coronary occurrences between men and women, and thus, of the application of "interventions" to reduce same.

Just today, I see a report in The Lancet, summarised on BBC, which included the following:
"Dr Maciej Tomaszewski, from the University of Leicester, and colleagues studied 3,233 biologically unrelated British men who were already enrolled in other medical studies investigating heart disease risk.

When they carried out genetic tests on the men they found that 90% possessed one of two common versions of Y chromosome - named haplogroup I and haplogroup R1b1b2.

And the risk of coronary artery disease among the men carrying the haplogroup I version was 50% higher than in other men."

Interesting as a piece of research, and suggestive of more research, but of no clinical value.

CarbSane said…
Thanks Sanjeev. I think everyone's in agreement that animal studies are a "last resort", but can be useful.
CarbSane said…
@Leon: Back when I worked in this area, the big controversy was over lack of drug testing in women. It is nice to see this has changed.

@Steve: Since you brought gender into the mix, DEPENDING on what is being studied, I may well put more credence into what happens in female rats than lean 20 y.o. men.

As with "class evidence" in forensics, lots of info in rodent studies can be of the exclusionary sort. For example, blood type is class evidence, blood evidence that could come from any number of sources sharing, say, type A. If the blood found at a scene is type AB, and a suspect is type A, we can exclude the suspect as the source of that blood, but if they are AB, we need more evidence to link the blood to them individually. There are a lot of theories out there and for some basics, animal models are highly useful in determining if it is even possible. Hope that makes sense.
Sanjeev said…
hey Steve

hypothetical for you: Suppose current cloning techniques were perfected, then hundreds of copies of you were cloned and used to test pharmaceuticals.

Would you apply the results to yourself unconditionally?
LeonRover said…
Oh, Sanjeev, a SteveC57BL/6 study ?

And how does one compare that with a LeonC57BL/6 study ?

Repeated tests on an individual can provide an estimate of measurement and dose variance. Population studies are a whole 'nother game, usually resulting in at least one order of magnitude increase in variance.

(While a student, I spent a Summer Vac analysing Latin Squares in an Agriculture Research Lab. It got old fast, as I was only given a 2-register desk calculator, paper and pencil. BUT I learned a lot about Fisher and other tests.)

All results are conditional.

More recently, I have concluded that I need 0.5 -1% confidence before accepting that results of a study might apply to me.
LeonRover said…
More recently, I have concluded that I need 99-99.5% confidence before accepting that results of a study might apply to me.
Sanjeev said…
It's actually a trick question. I'll wait if anyone else wants to chime in before giving my take ...

I bet you haven't felt this much suspense since we saw the "I love Genie" guy fall out the window on "Dallas", or Gandalf got killed in the 1st movie ; )
Lerner said…
how many avenues of investigation that look promising in rodents fail to pan out in humans?

Now for Pharma, I believe figures show IIRC that for every ~1000 compounds that are investigated, only 1 makes it to market. So how many rat promises also fail?
Lerner said…
with a thousand clones, they still won't necessarily be in the same state as me. E.g., I find myself being sensitive to whey again (burning eyes). I did have a GI virus not long ago, so I'm guessing some gut permeability is present (still).
CarbSane said…
I'd venture to guess that the percentages of drugs shown effective in rats that are effective in humans is actually rather high. The reason for the statistics on candidates making it to market are governed by a lot of other things. Now, granted, this was over 20 years ago, but here's how it worked when I was in the field. A general structure is pharmacologically active. The organic chemist synthesizes a bunch of similar structures changing a functional group here or there, switching out a bromine for a chlorine, etc. These compounds are first put through in vitro testing for overt toxic or mutagenic activity. If they pass, then they are put through a cursory toxicity test. Pass that and then they are tested broadly for efficacy in rodent models. If there were signs of efficacy, I would get the drug -- compound # such and such -- and my job was to (a) develop the assay to quantify blood levels, etc. and (b) perform bioavailability, pharmacokinetic, and tests on the metabolism/fate of the drugs in rodents. I would estimate that I worked on perhaps 50 or so such compounds in 5 years. We would also get samples from toxicology to analyze as the process was begun to determine dosages and tolerability and from another department that determined therapeutic dosage. I don't recall that even one of my discovery level drugs ever made it to the next level. It's possible one or two did, but I was long gone before our department would next be called upon. I don't know what the rate of drop-off is from a drug that has been named, extensively tested in animals and shown effective once it gets to humans for clinical trials, but I doubt the number that are totally ineffective in humans after working in animals is that large -- often higher animals including monkeys are tested first as well. A lot of drugs die in clinical trials but mostly this is due to unforeseen side effects, unacceptable risks and such rather than it not working at all.

The diet world is a bit jaded by the leptin story. But this is because the ob/ob mouse is leptin deficient and we now know this is just not the case in human obesity to any large extent.
Sanjeev said…
Lerner and Leon, here's the trick:
clones are not your identical twin.

All current technologies would yield you BUT with someone else's mitochondrial DNA.

So one could theoretically still get killed by an agent that cures a disease in one's clones.
Lerner said…
that's clever, Sanjeev - and most (but not all) of my clones agree :)