Wednesday, November 27, 2019

Of mice and men

I usually start from the premise that insulin makes you fat. The most simplistic prediction from this is that eating carbohydrate raises insulin and this insulin is what makes you fat. Over the years I have looked at data from all sorts of places, particularly the extremes such as the Kempner Rice Diet and/or the Potato Diet, which clearly work and which appear to do so (to me!) via lowering the level of systemic insulin acting on adipocytes. In particular I consider that first pass hepatic insulin extraction has a huge effect on the systemic insulin level and the subsequent exposure of adipocytes to that insulin. The carbohydrate-insulin-model of obesity as set up by detractors is simplistic in the extreme.

This paper was published recently:

The carbohydrate-insulin model does not explain the impact of varying dietary macronutrients on body weight and adiposity of mice

Of all of the combinations tested the one which we are interested in is where protein was held constant at 10% of calories and carbohydrate was varied from 10% of calories up to 80% of calories. The remaining calories were made up of fat giving a pure carb vs fat comparison. At the end of the study period we have this graph where I have added in the percentages of calories from fat in red along the x axis:






















If you draw a straight line through the data points your weight line slopes downwards as fat percentage drops, pretty well. Fat makes you fat:





















Now, clearly, there is a missing data point. That is the bodyweight from a diet group with zero carbohydrate, 10% protein and 90% fat. This combination was not included in the study.

So all else from here forward in this post is now pure speculation.

Help is at hand for the missing data point in this paper

A high-fat, ketogenic diet induces a unique metabolic state in mice

Here we also have C57Bl/6 mice and in this case they were fed F3666 diet which looks like this:

"The proportion of calories deriving from different nutrients was as follows: ... KD:95% fat, 0% carbohydrate (0% sucrose), 5% protein"

This is not a perfect fit as the protein is about half that used by Speakman and there is (obviously) no sucrose, but it's the best anyone can do in the absence of the omitted group essential to complete the graph. The mice which were eventually put on to F3666 were initially made obese with a sucrose/fat combination before being put on to their ketogenic diet. Their weight dropped from approx 37g on the sucrose/fat diet to approx 27g on F3666, ie they ended up about 3g lower in total bodyweight than the control mice fed approx 10% of calories from fat in a carbohydrate based diet throughout. Here's the graph we all know from years ago. Open triangles show the drop in weight when F3666 was introduced at around day 80:
















So what might a zero carb, fairly low protein group of C57Bl/6 mice look like? They might well end up slightly below the weight of Speakman's 80% carb group mice, (ie those eating closest to mouse chow), or they might end up slightly heavier due to the higher protein content limiting the total fat percentage which could be provided. I feel a compromise might be to use 35g, the same as Speakman's 80% carb group, the closest chow equivalent.

I've added the zero carb speculative data point to the blue line on the graph at 35g bodyweight which now looks like this:






















and now we can curve fit the bodyweights like this:






















Ah, that's better. Ketosis at the left, "carbosis" at the right. Nice.

I love rodent studies. You just have to understand that setting them up correctly is essential to obtaining the result you want. You also have to know what you want.

Peter

30 comments:

Tucker Goodrich said...

Did you look I to the n-6 content of Speakerman's diet? F3666 contains quite a lot, which should help the mice lose weight as it fattifies their livers...

Peter said...

Yes, first thing. The fat mix is about 11% omega 6 PUFA. Even at 80% of calories from fat that's only around 8% PUFA. And the very high fat diets clearly had the highest PUFA but that's were the downward trend in bodyweights began. I don't think 8% PUFA is a huge problem for Bl/6 mice, especially if there is a preponderance of saturated fats over MUFA in the mix, as there appears to be here.

Re F3666 it's hard to say quite what they fed. BioServe do specifically make it choline deficient which means the fat can't be exported from the liver unless as FFAs or ketones. But the paper says that they asked BioServe to add M-vits of some sort to the F3666, so who knows. There is a related study

https://www.ncbi.nlm.nih.gov/pubmed/24009777 and also

https://www.ncbi.nlm.nih.gov/pubmed/29045796 which probably added choline

F3666 is a very lose canon!

Peter

Tucker Goodrich said...

Interesting. 8% n-6 was where Alvheim et al got their bang in making C57BL/6j mice fat. I pointed that study out to Speakerman on twitter back in January, probably too late to have any impact on this study.

"The Alvheim experiments used 1 & 8% E LA to model the increase in the human diet, in the 60%, 35%, and 12.5% fat arms.

"If you replicated that with this intermittent HF diet you should show a lower effect in the 8% E LA arms, as LA would average 4.5%...

"Add a keto arm!"

https://twitter.com/TuckerGoodrich/status/1082800943220092928?s=20

I haven't been through this new one yet though. I still don't really buy the CIM absent sufficient n-6 to induce IR and the rest; and this, from what you say, doesn't increase my confidence in a non-n-6 CIM obesity hypothesis at all.

Tucker Goodrich said...

Whoops, *Speakman!

raphi said...

I asked Speakman on Twitter, lets see :)

Puddleg said...

It's beyond plausible that LA, and specifically the SFA/LA ratio in the presence of whatever omega-3 there may or may not be, is the missing precondition that explains the near-universality of the CIM in modern humans.
See for example
https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6340377/

https://onlinelibrary.wiley.com/action/showCitFormats?doi=10.1002%2Fmnfr.200900516

If the CIM is a hypothesis that needs modifying, the omega 3/6 scientists have been quietly modifying it for years without causing any great fuss, and it's time this was recognised.

Ginger's Keto In The UK said...

dear Peter, finding out about your theory of Reverse Electron Transport thorugh dr Eades gave me the proverbial hidded variable in obeservation I did for the 6 years I have followed and tested with the ketogenic diet. I am doing my best to find more sources and propose my end of year essay (I study biosciences) on the subject but I am strggling to find papers and studies that have the metabolic take on the reverse electron transport beyond the simple beta oxidation, and all seem to just mention the negative effect of ROS to contribute against anything that generates them. If you could point me to any resources to star research for my bibliography i would be immensely grateful forever.
you can also reach me at ginger@ketointhe.uk

thank you so much

ctviggen said...

Thanks, George, for this study:

https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6340377/

It's quite enlightening.

Peter said...

Hi Puddleg,

I really love that first ref. Take a drug which makes you hungry, this down regulates the hunger receptors, you then have reduced hunger longer term.

Would it work with depression? Could you, in an opioid naive patient, give occasional doses of naloxone opioid blockade with the view of up regulating the sensitivity of endorphin receptors so as to allow basal endorphins to produce happiness? I’m not thinking of continuous naloxone use or the long acting naltrexone. Maybe a single naloxone dose giving 15 minutes a day of naloxone induced endorphin blockade once a day? Once a week? To up regulate the receptor sensitivity for happiness. The opposite of addiction where transient happiness has to be paid for by a desensitised happiness receptor? Of course giving an antagonist might not have the unlimited ability to hypersensitise in quite the same way as an agonist carries near infinite ability to desensitise…..

Sorry to rabbit on, that one rattled my chain rather! Been thinking about it for years.

Oh, penetration of insulin past the liver to hit peripheral adipocytes: Anything which causes lipolysis in the presence of dietary carbohydrate will cause hepatic insulin resistance and so reduced first pass insulin extraction. Systemic fructose, alcohol, nicotine to start with. I have long wondered if trans fatty acids block fatty acid negative feedback so cause inappropriate lipolysis. Possibly growth hormone but that one is a little problematic. Caffeine looked to be on the list but is interesting in how it does what it does, does not seem to be in the same problem area as nicotine.

I’ve never really thought about the n-3 n-6 ratio other than as a surrogate for avoiding omega 6s. Might have some myopia there!

Peter

Peter said...

Hi Ginger, I think I started here

https://www.ncbi.nlm.nih.gov/pubmed/23730255

and here

https://www.ncbi.nlm.nih.gov/pubmed/17919343

but there is a huge supply of information in Pubmed which will be more up to date but, equally, more lost in detail available today. See if those gets you started.

Peter

Peter said...

raphi, yes, it will be interesting to see if anything is said. I think the correct approach would be to stick with his data and leave the speculation to outsiders!

Peter

Gretchen said...

It would help if graphs had labels that explained what ab, bd, etc. mean.

Unknown said...

"90% fat & 10% protein" -- we just did that study. Submitting manuscript soon. Interesting [discordant] effects wrt GTT, ITT, ISI.

Gyan said...

Is the zero-carb data point strictly necessary?
The first figure itself has a peak at 30% C and to just plot a straight line-how do they justify it?

Gyan said...

It again seems to come back to the old idea that the F/C ratio of ~1 is esp unsafe and one should eat either high fat or high carb.

Now, does the SFA/PUFA ratio matter in context of a diet with F/C ~1. That is where it gets personally interesting to me.

Alex said...

@ Ginger's keto in the UK

You might want to look into the following: "Glucose restriction extends Caenorhabditis elegans life span by inducing mitochondrial respiration and increasing oxidative stress."

Whether ROS does damage or not depends rather on the location: mitochondrial (fat) vs glucose (mito + cyto). oxStress is no problem if the mitos adapt to it via increasing anti-oxidative systems = oxidative resistance goes up. i think with glucose there is no mitochondrial adaptation: ROS not so good long term.

Also, you might look into exercise studies, which showed a lack of adaptation with vit E and C supplementation. (eg PMID: 22928084 ... take it from here).

Good luck.

Unknown said...

Thanks for taking the time to blog our article. It's a shame you took half of the space discussing a data point that didn't actually exist. Especially since you could have fitted the curve you ended up fitting to the actual data in the paper and would have got something virtually identical.

You may be interested to know that partly prompted by comments on twitter since we completed this study and published the first paper from it last summer we also did some additional studies where we have pushed beyond the boundaries defined by the 24 diet matrix. This includes a set of diets with 10% protein and between 80 and 90% fat. Under analysis at the moment. This will complete the 'ketotic' end of the diet spectrum.

As regards the fat composition of the diet. We aimed to do 3 things with this. The first was to have an absolutely fixed fatty acid composition independent of how much total fat there was. This was led by the fact that the most popular diet series used in mouse studies is a mess of confounding changes in composition. The second point was to make the SAT:MUFA:PUFA ratio match that in the standard american diet, and the third point was to have the same n-3:n-6 ratio as in the american diet as well.

I have been pretty surprised to find that proponents of the CIM have criticised this diet saying that the CIM isnt expected to work when there is a lot of saturated fat in the fat component. I dont know where that comes from but it is a pretty fatal problem with the CIM because effectively it means as a model for why we get fat it doesnt apply to vast swathes of the US public.

best wishes
John Speakman

Alex said...

@ Unknown/J. Speakman

@1st paragraph: You initiate your comment with condescension and cynicism. That's not cool. It's unprofessional.

@2nd paragraph: You are using a sarcastic tone to point Peter towards unpublished data. (Pete, you should know about data they have not analyzed yet!) That's not cool, Speakman!

@ 3rd paragraph: are any of these arguments in your paper? No. (your keywords not found in paper, the word omega is not to be found...) Why not? Why do you need to justify your study in a blog? How could Pete, or anyone else, give you credit and discuss the results without knowing your rationale?

@4rd paragraph... well, you are in the right place ...

Best regards
Alex

Unknown said...

Dear Alex

The first paragraph just states a simple fact. Half the blog is about a data point that doesn't exist, and he could have come to the same conclusion just fitting a curve to the data that do exist. why is that unprofessional? how dod you conclude it is written with condescension and cynicism?

In the second paragraph I mentioned the work underway because in an earlier post he had mentioned asking that i do such studies. is that really written with a sarcastic tone?how do you even do such a thing?

Regarding the third paragraph The complete rationale for the study was already published in Hu et al 2018 Cell mMtabolism. I was reiterating it here because he may not have seen it in the previous paper. Journals tend not to like you repeating stuff from previous papers because of copyright, which is why it isnt in the Molecular metabolism paper.

Hope that clarifies some things. I dont NEED to justify this study in a blog. Just since I did it I am probably best placed to provide some answers to the questions raised. I'm sorry if you feel threatened by that.

best wishes, John.

Peter said...

Dear Dr Speakman,

There have been many criticisms on twitter of your study with which I would disagree. In particular I feel that the mitochondria of mice have quite a lot in common with those of humans (with a possible exception of C57Bl/6 mice which have several quite unique metabolic problems. Happily they seem to be behaving pretty well here), so I am always interested in studies like this. My interest is primarily is what stops adipocytes accumulating lipid in the aftermath of a meal. There appears to be an interaction between fat, fat saturation, insulin and insulin resistance which predicts very much the curve you have produced but with the most marked weight lowering effect present at the highest fat intakes.

My huge openly declared bias is in favour of carbohydrate restriction so I particularly look at this end of any data published. Ultra low fat diets are ubiquitous in mouse studies (and limit weight gain) so the interesting data points are those at the very low carbohydrate end where weight gain should be equally or more limited than at low fat intake.

My second bias is in favour of saturated fats so your use of a significantly saturated diet compared to many in mouse studies is a plus. The inclusion of increasing amounts of polyunsaturated fats is a potential confounder in many studies. My view is that the more double bonds in a fat molecule the harder it is for an adipocyte to limit fat ingress. Although, as Tucker mentioned above, as little as 8% of total calories from omega six PUFA can be shown to be obesogenic in mice my own feeling is that this is much less likely to occur if the background fat is higher in saturates rather than starches. It’s a long story.

Data from carbohydrate restricted, high saturated fat diets are like hen’s teeth.

The genuine data point from near ketosis will be of great interest when published.

Thank you for your comments.

Peter

Peter said...

Re the soon to be published data on 90% fat 10% protein diet: These mice should be 1) glucose intolerant on IPGTT, 2) Insulin sensitive on exogenous ITT and 3) It would make my day if the nadir of blood glucose during ITT was delayed by about half an hour.

Fingers crossed.

Peter

Peter said...

Alex, there are some useful data coming. Let’s wait and see.

Gretchen, it’s free full text so you could check but I guess these are just statistical significance markers.

Gyan, it’s biology, there is often a lot of spread. But I agree, it was the curve the caught my eye initially. It’s even more marked in the 25% protein group. By the time you are getting in to 50:50 calories the PUFA content is getting pretty low and overall saturates are quite high here so we are probably seeing a real effect (if the curve is genuine) without PUFA induced obesity to mess it up. So yes, F:C ratio might really matter but it shouldn't be linear…

Peter

Gyan said...

Peter,
I should have thought that 80% fat diet is pretty ketogenic. Why do you expect some special effect at zero-carb?.

The authors do not consider a zero-fat diet as well so they can hardly be blamed for ignoring the zero-carb diet.

Peter said...

Gyan, mice are mice and have very little brain to support so they are not very good at ketosis. I pulled a random D’Agostino paper and they used 88% fat from almost totally MCT oil, 11% protein and zero carbs. Even with this level of effort they only achieved BHB levels around 1mmol/l at 4 weeks dropping to 0.5mmol/l by 16 weeks. But this doesn’t matter, what matters is input at ETFdh in adipocytes. Ketones themselves only appear to input at complex II, either via acetyl-CoA turning the TCA or via the conversion of succinylcholine-CoA to succinate for complex II as acetoacetate is converted to acetoacetyl-CoA. Enough saturated fat should limit insulin’s action adipocytes via ETFdh. Ketones are less interesting except as a surrogate for elevated FFAs, hopefully stearate and palmitate.

Peter

Peter said...

That will be ketones only input as FADH2 at... etc

Peter

Passthecream said...

Peter: "Take a drug which makes you hungry, this down regulates the hunger receptors, you then have reduced hunger longer term."

A sort of chemical hardening.

It reminded me of something - one of my colleagues keeps getting pulled over while driving because traffic police think he's talking on his mobile phone but what he is actually doing is angrily shouting at his radio whenever politicians are being interviewed. That seems like a great way to de-sensitize, after listening to them anything else seems normal.

Peter said...

Even on the radio you know that their lips are moving at the mike. Says it all. Certainly about our Boris.

Peter

Brad Marshall said...

Hey Peter!

I've just completed an N=1 test of the protons theory. I made a croissant based diet based on the stearic acid diet given to the mice in Valerie Reeves thesis and it worked!

Details at https://fireinabottle.net/introducing-the-croissant-diet/

Thanks!
Brad

karl said...

Grumble mumble - groan..

I have a real problem with these 'test diets'. I just don't think this rises to real science. At the very least, one would have to start with a feed lot and add a single thing as a variable. But that is not what happens. If you get a test diet from one year - is it the same as what they mix up the next year? Different growing seasons - different fields - different fertilizers - a bunch of confounding variables.

I think we know better than this - in real science we should expect that there is a SINGLE variable. Fat is not a 'thing' (fat is a class of thingS ) SFA, MUFA PUFA are classes of things. Why did science go off the rails?

Back in the 1960's there was work on synthetic chemically defined diets - they really didn't want to deal with astronaut poop.. The base of this work was by Winitiz, Greenstien Seedman, Graff - others that I don't remember.

To figure this stuff out, we need a single chemical variable.. - such as the amount of LA (where there is only one naturally occurring isomer).

If someone had been paying attention, this early work showed that fructose spiked trygly where glucose didn't - and should have put an early end to the cholesterol nonsense:

https://www.sciencedirect.com/science/article/abs/pii/0003986164904461?via%3Dihub

Why waste money on no chemically defined diet studies? Some chemically defined diets are still around - https://www.nestlehealthscience.us/brands/vivonex/vivonex-plus-hcp

ctviggen said...

Brad, that's a very interesting N=1 study. Thank you for undertaking that.