Here I come to save the day ....

.... it means that Mighty (Meta)Mouse is on {cue record scratch sound}

A couple of years ago I wrote the following blog post:  Mighty Metabolism Mouse discussing this study:  A high-fat, ketogenic diet induces a unique metabolic state in mice.  

If you take mice and put them on a low protein (5%), very low carb (2%) ketogenic diet, they lose some body weight, despite consuming the same number of calories.  Taubes' theories were already waning in popularity by at the time of that post, and folks were hanging their "metabolic advantage" hats on studies such as this one.  It turns out that they lost weight to the tune of what mice consuming a calorie restricted diet (only 66% of baseline calories, or 1/3rd less) did, and had energy expenditures roughly 11% higher than the regular mice and 15% higher than the calorie restricted ones.  In other words, they don't violate the first law of thermo or the common dietary acronym CICO, but this proved that ketogenic diets increase calories out so that these mice effectively could eat more.

This is why I dubbed the KD fed mouse the Mighty Meta Mouse.  Now, this superpowered metabolism was not without some inconvenient side effects, most disturbingly hepatic steatosis, aka "fatty liver".  But from a weight loss angle, a closer look revealed that this keto diet wasn't all it was cracked up to be in that regard either!  As I've highlighted previously, the low insulin state is associated with fuel partitioning to fat at the expense of lean tissue, not the other way around, as we're told by the insulin is *the* fattening hormone crowd.  The keto mice had more fat and less lean than the chow fed mice despite having one-tenth the basal insulin!!

I followed up that post with some musings on the applicability to humans.  I cannot stress enough how rodents differ from humans, between species of rodents, and even between different strains of the species.  We need to be very careful when extrapolating the results of such studies to humans.  HOWEVER .... Time and again, the rebel-without-a-clue "alternate" nutrition community repeats the same transgression:  

• Study seems to support your diet, LOOK HERE!!!   Diet not quite yours?  That's OK, close enough.  Pigs are humans right?  Everything your doctor tells you is wrong ... you know the drill.
• Study seems to refute your miracle diet, DON'T LOOK THERE!  Good thing I'm not a rat (or mouse or pig), that diet is not really like my magic diet, it contains more carbs or the wrong kind of fat, etc.  ... you know that drill too.

You cannot have it both ways.  To me, the moral of this particular experiment was that "unique metabolisms" induced by extreme diets can be created at great expense to the organism.  Would this "expense" be paid by the human?  There's a reason we'll never know for certain.  

But I've been banging one drum about mice for a while vis a vis the whole "insulin is the fattening hormone" thing, and yet another (sigh) book invoking "science" to show how insulin causes us to cannibalize the lean and give to the fat.  Even more simply put, "insulin is the fattening hormone" doesn't hold water in the face of massive amounts of rodent data.  Unless you're going to claim that insulin does different things to the mouse than to you and me (careful, if you do you might as well toss out all that elegans research! ).  The most cognitive dissonance inducing mouse is LIRKO.  That mouse, lacking insulin receptors in the liver, becomes profoundly hyperglycemic and hyperinsulinemic but not obese or diabetic.  Indeed the longer things go for this mouse it becomes hypOglycemic while it's hulk of a  pancreas keeps pumping out insulin ... and still no obesity.  In our Meta Mouse, we have almost no insulin!  And yet, while the standard fed mice averaged 33.5 grams body mass, only 13.5% of that was fat, Meta Mouse only tips the scale at a svelte 29.0 grams!  Yay!!   Errrr ... not so fast, because MM is also 17.8% body fat.  For those into statistical fun and games, that's almost 14% more fat mass in absolute grams, but a whopping increase of 31% of the body fat percentage of a normally fed mouse.  It's just not looking good for the "lower my insulin" gang at the moment.

Oh wait!  These are mice!!  I almost forgot.  Someone remind Fat Head, because if calorie restriction making mice fattier is blogworthy, then surely ketogenic diets doing the same is at least worthy of comment {chirp chirp}.  

Now the other day on Facebook, Jimmy Moore posted this in response to a new study on a ketogenic diet in mice:

Here's the PubMed Link.  The responses in comments were typical, a bit amusing, and at the same time quite annoying.  To paraphrase the worst:  

How can the mice become diabetic if they eat no carbs.  This makes no sense!

You're right.  That makes no sense if you have a simplistic view of diabetes as fed to you by the irresponsible advocates of low carb "science".  To be fair, this idea that carbs cause diabetes has been somewhat adopted in the mainstream,  but this does not excuse the so-called geeks and buffs who claim to regularly research this stuff.     First of all, while there is SOME evidence that ketogenic diets are a treatment for diabetes, there is no evidence, that I'm aware of, that ketogenic diets are preventative for metabolic syndrome or diabetes.  As a treatment for T2, keto almost never seems to improve -- either anecdotally or experimetally -- insulin action in response to a glucose load.  Let's be clear here as well.  We're talking ketogenic diets, not low carb, most of which are not ketogenic after several weeks.   How many "my glucose goes over 200 when I eat a small potato" accounts do we need from formerly non-diabetic low carbers before they wake up?  

There is very limited human data on this.  The rodent data does NOT look promising.  For all of the "get off the meds" claims made about low carb diets, the longest study in diabetics didn't really deliver quite what it was hyped to.  It seems that "resting" your pancreas isn't doing it much good after all.    

Before I delve into this new study, it seems that Richard Feinman had a "me too" moment about keto mice that don't lose weight.  Heck, he even managed to make them obese!!  Paging Fred "you can't get obese on LC" Hahn!!!  Feinman tweeted out a link to this study on Pinterest:  Response of C57Bl/6 mice to a carbohydrate-free diet.   The chart of the weights of these mice shows a different story vs. Kennedy (the Meta Mouse).   The difference? Perhaps because Feinman's mice consumed a higher protein (20%) chow than did Kennedy's (5%).  So ... maybe protein really is like chocolate cake after all, because while initially they consumed about 10% more calories, this leveled out after 4 weeks and they ate comparable amounts.

WHAT?  You read that right, in this study, as compared to the Kennedy study, the keto mice ate roughly the same amount of calories but just got bigger and bigger and appeared to be getting bigger still!  Feinman and co-author's conclusion?  

The results suggest that rodent models of obesity may be most valuable in the understanding of how metabolic mechanisms can work in ways different from the effect in humans.

Ya got that??  Get a result different than expected, or different than other researchers got (that bolstered your views) and now rodents become useful in elucidating our differences.   Because you see in mice:

Consistent with the obese phenotype, experimental mice had fatty livers and hearts as well as large fat deposits in the abdomino-pelvic cavity, and showed impaired glucose clearance after intraperitoneal injection.

Worry not though, because ...

The high fat-fed mouse is a widely investigated model of obesity, insulin-resistance and susceptibility to diabetes. The C57Bl/6 strain, in particular, when subjected to high dietary fat, shows increased consumption, increased efficiency of fat storage (weight gained/calorie) and becomes insulin resistant [1-4]. As a model of human obesity and insulin resistance, however, it suffers from the severe, if under-emphasized, limitation that high-fat diets do not generally cause these conditions in humans unless the diets are also high in carbohydrate.[No Citation Here]

That last paragraph is from the Introduction, so as we can see, there's no scientific bias going in here.  Regarding the C57Bl/6, the impression is given that these were bred to be particularly susceptive to obesity, something I found no evidence for in the literature last time I looked.   They will fatten in response to a "fattening" diet, however.  Here's the JAX low down on our ketoschizo mouse.  Here's the concluding paragraph from reference 4:

Our findings suggested that when mice were fed an HFC diet, they could develop initial compensatory response to resist the increased energy balance; however, a prolonged consumption of an HFC diet appeared to disrupt this adaptation.

It is interesting to see Feinman -- in peer review print -- try to rationalize away the very same thing he would use to justify his insistence that a calorie is not a calorie.  Remember that debacle of a paper?   Feinman (and his cohort in thermodynamics-mangling crime, Fine) relied heavily on two things for their metabolic advantage:

1. The metabolic cost of gluconeogenesis, and
2. Thermogenic differences between macros as ascertained by Eric Jequier.  

Regarding #1, is Feinman now saying that gluconeogenesis is not energetically costly?  I know mice require less glucose for their tiny brains but they still must make it if they are on an essentially zero carb diet.    But surely someone as familiar with his work as Feinman has seen Jequier's Pathways to Obesity paper.  I blogged on that here.   The image at right from the blog post and F&F's paper shows their progression of effective calories for 2000 calories -- beginning with 55% carb, 15% protein and 30% fat, then substituting 1:1 protein and fat for carb.  Leaving aside their math error, they say that at 21% carb (47% fat, 32% protein) you have a 100 calorie deficit in net calories after TEF.   

This is disingenuous because the only way reducing carbs is TEFtacularly metabolically advantageous is to use protein in the substitution.  Because if you just replaced the 34% of carb in the above experiment with fat, you would end up with about 30 more calories of net energy.   For the mouse, that usually gets about 80% calories from carb and replace that with fat.  For every 100 calories of carb replaced with fat, you get 4.5 more net calories from the fat.  But their KD chow was higher in protein.  But ... but ... Oh yeah ... that doesn't count in mice though ..... (Long time readers are probably aware that my view is that TEF is likely irrelevant for a warm blooded animal, as that is just that much less energy that must be used elsewhere to generate heat).

Must be something ELSE going on.

OK ... Finally, let's get to that new paper:

Long-term ketogenic diet causes glucose intolerance and reduced beta and alpha cell mass but no weight loss in mice.

They used a keto diet (PDF download) from Research Labs.  Now, some are hanging their hats on this having corn oil in it (works out to just under 12% total energy).  The rest is Primex(?) which I cannot find any info on.  Perhaps someone in this community that works for this establishment might know ;-)   Here's how the macros are reported in the paper.

In any case, they took 10 week old mice (so they'd been eating the standard low fat -- usually 10-15% fat -- for about 6 weeks at the start) and put them on KD or standard chow.  The dark circles are KD, and they initially dropped weight, but recovered and matched their controls by about the 10 week mark.   So were there other differences?  Yes:

Our data show that long-term KD causes dyslipidemia, a pro-inflammatory state, signs of hepatic steatosis, glucose intolerance and a reduction in beta and alpha cell mass, but no weight loss.

Boy am I glad I'm not a mouse eating corn oil and Primex!!!   Nothing to see here.  Not so fast!  I looked into this aspect of things a while ago as well.  Here:  Ketogenic Diets & Fatty Liver.  Be forewarned, that I get a bit snarky at Peter Dobromylskyj aka Hyperlipid in that post.  In my opinion his advocacy of high fat diets crossed a line when he began openly admitting to manipulating data to fit his biases. The acronym CIAB is pretty well known in these circles because of him, standing for Crap In A Bag.    Peter will write off any inconvenient results when the diet contains PUFA or transfats, and yet when other studies use those same fats and get positive results for his view, that's ignored, as are studies that yield negative results using lard or butter.   As stated in the post, when one casts the net a bit wider we find that butter is actually quite "toxic" and fatty-liver promoting ... and INFLAMMATORY.    Peter has also "speculated" about how a fatty liver might even be beneficial, and is likely responsible for people around the IHC referring to "physiological insulin resistance" as something beneficial.  Heck, some have followed his lead so far as to imply it is life saving (but only if Robert Lustig is chasing you with a syringe full of insulin and you are not diabetic).

What made this study unique is that they investigated what happened to the pancreas ... and not just the beta (β) cells (that produce insulin) but also the alpha (α) cells (that produce glucagon).  In most diabetics, it is the β-cells that are compromised (in T1's completely) while the α-cells remain intact.  Glucagon's major role is to prevent hypoglycemia by stimulating endogenous glucose production, EGP, in the liver, a combination of glycogenolysis (glycogen breakdown) and gluconeogenesis (synthesizing glucose from glycerol, certain SCFA or certain amino acids).  Contrary to low carb lore, if protein "spikes" your blood sugar, it is because of an inadequate insulin response to glucose to suppress glucagon.  (The Cause of Hyperglycemia in Type 2 Diabetes).  

So:

• Triglyceride content of the liver was more than 2X controls
• Circulating lipids, hormones, etc.:

• I would note that in this case (vs. Kennedy) circulating triglycerides were also elevated in KD.
• The progression of glucose tolerance.  Early on, there was reactionary insulin hypersecretion to maintain normoglycemia.  By 5 weeks there is a worsening of glucose tolerance that is described as "profound" by the 12th week.  Note how the early peak of insulin (though not pronounced in controls seems to be muted in the KD).   It's important to note not only insulin levels but changes and the impact on glucose levels.  At 20 weeks, the mice appear to be hyperinsulinemic in the fasted state.

• "After 22 weeks beta cell mass in KD-fed mice ... was decreased. The density of islets was unchanged, but the median beta cell cluster size was significantly decreased. This was due to a relatively increased number of islets with a size smaller than 2500 μm2 in KD-fed mice."
• "The alpha cell mass was reduced by 50% in KD-fed mice after 22 weeks"
• Taken together the ratio of alpha to beta cells decreased

"This indicates that long-term high-fat, low-carbohydrate ketogenic diets lead to features that are also associated with the metabolic syndrome and an increased risk for type 2 diabetes in humans."

Interestingly, these researchers did address the fatty acid content of the diet:

Dyslipidemia and the pro-inflammatory state induced by KD may be the consequence of the high content of saturated fatty acids. It was shown that a short-term polyunsaturated fat enriched KD did not adversely alter lipid metabolism in adults (15). Also supplementation of a high-fat diet with omega-3 polyunsaturated fatty acids has been shown to prevent high-fat diet-induced insulin resistance by reducing inflammasome-dependent inflammation in rodents (41). However, in our study the higher content of omega-3 fatty acids in the KD could not prevent pro-inflammatory effects on the long-term. Whether further modification of the fatty acid content of KDs can attenuate dyslipidemia and the pro-inflammatory effects on the long term needs further study.

What about choline?  Addressed:

However, the similar methionine and higher choline content of KD in our study did not prevent signs of steatohepatitis on the long term. 

There were two other major points in the discussion (without repeating the obvious).  First, that their study was longer and in somewhat older mice than previous studies.  Short term results were not predictive of the long term effect of a diet.  Keep this in mind when human studies on keto and LC diets last in the 3 to 6 month range.  It is quite possible that some of the trajectory of weight rebound is metabolic adaptation as seen in the mice.  But this is a stark reminder that short term improvements cannot be translated to the long term.   Here is where I always get myself in trouble, but I can't help but point this out.  The ONLY long term human data we have is anecdotes from LC advocates, yet another one of whom passed away at age 71 "unexpectedly" of undisclosed causes last month.   We're told all the time -- and I agree to a large degree -- that we can't always tell a person's state of health by looking at them.  All of these healthy looking low fatties are really just one Oreo away from dying ... don't let appearances fool you!!   It's the low carbers who look generally less robustly healthy than their peers ....   (sigh)

The second discussion point is in regards to the alpha cell demise.  They write:

In relative terms this decreased alpha cell mass was even more pronounced than the reduction in beta cell mass resulting in a major decrease of the alpha-to-beta cell ratio. Decreased glucagon levels lead to less gluconeogenesis from the liver, which may prevent hyperglycemia in KD-fed mice.  Whether this change in alpha cell mass is a direct consequence of KD or a response to counteract glucose intolerance remains to be elucidated.

These mice did not have elevated fasting glucose and it would appear to be due to reduced glucagon levels.    I wonder what some of these long term low carbers have for glucagon levels?

EDIT:  I'm going to add this one more study (ht Charles Grashow) so they are all in one place.  A high-fat, ketogenic diet causes hepatic insulin resistance in mice, despite increasing energy expenditure and preventing weight gain.  Similar story:  Increased hepatic lipid accumulation, hepatic IR and diminished whole body glucose disposal.  /EDIT

In Summary:

• Ketogenic diets often result in short term weight loss but over longer terms they can even result in obesity.
• KD have a propensity to produce hepatic steatosis
• KD often result in most/all features of metabolic syndrome
• KD appear to reduce pancreatic function

... in mice.  I'll leave you with two more old studies/blog posts of mine:  How Fatty Diets Cause Diabetes, and Revisiting the Fatty Diets & Diabetes Study ~ How to Make Mickey Fat or Fattier.

What do ketogenic diets do in humans over the long term?  Do YOU want to be a guinea pig??