Wednesday, December 12, 2018

A post not about Walter Kempner

I've had this paper on my hard drive for a while. It's been sitting somewhere near the front of the back of my mind but was doing nothing to really grab my interest.

Chronic high-sucrose diet increases fibroblast growth factor 21 production and energy expenditure in mice

I've got a draft of a post from mid summer this year which I wrote simply because I like the attitude of the authors. They say things like:

"Excess carbohydrate intake causes obesity in humans".

That's the first line of the abstract. You know, it's that "nailing your colours to the mast" sort of a statement. Even though I do think life is a little more complex than that.

Anyway, I like these folks who are looking at the slimming effect of sucrose in BL6 mice. That's correct, sucrose is a slimming drug/food in mice, under the correct circumstances. People too. The data in the 2017 paper is an extension of the work they did in 2012, written up in this paper:

Ingestion of a moderate high‐sucrose diet results in glucose intolerance with reduced liver glucokinase activity and impaired glucagon‐like peptide‐1 secretion

I don't intend to go through either paper in detail, it's just that the 2012 paper has some rather special macro ratios that caught my eye.

This is what they did to the mice in that original paper:

"After adaptation for 2 weeks, they [the mice] were divided into three groups and fed a normal chow diet (NC), a high‐starch diet (ST) supplemented with 38.5% corn starch or a SUC containing 38.5% sucrose; the latter two diets were prepared by the addition of corn starch or sucrose, respectively, to CE‐2 (Table 1)"

Essentially they are diluting chow with starch or sucrose. Here is Table 1 for the diet compositions, note my red rectangle:

With group sizes of n=4 and five weeks on the diet very little of anything reached statistical or biological significance. The 2017 study used a slightly modified version of the diet to keep a low fat percentage identical across the diets but still had 38.5% of calories from sucrose, was run for 15 weeks and had group sizes of n=8-10. Results were statistically significant all over the place and suggest that the sucrose diet is decidedly good for metabolic health and gives a slim phenotype on ad lib consumption. Just so long as fat calories are very, very low. This looks very much like what Denise Minger described as carbosis, based in part around Walter Kempner's very effective, very unpleasant, ultra low fat, high sucrose medical diet. The Rice Diet is very real.

This post is not about any of the above.

Now, watch carefully. I'm going to sneak in some more macros

If you wanted a "reduced" fat diet which induces carbosis in human beings I recon the red text is pretty well it. I particularly enjoyed that exactly 7.7% of calories came from fat in each diet, this could almost be deliberate. If you combine what is almost certainly a very effective spontaneous weight loss diet with a 30% calorie restriction I suspect you might be on to a winner when comparing it against a reduced carbohydrate diet. Of course to really nail it you would have to compare it to an absolutely non ketogenic diet, say one supplying a total of 140g/d of carbohydrate. Does carbosis beat a middling carbohydrate mixed diet? You bet.

Oh, the scribbled-in red numbers came from Table 2 of this paper.

Most people in respectable CICO based mainstream nutrition have never heard of carbosis, Walter Kempner, the Rice Diet and have probably never heard of Denise Minger.

But Kevin Hall has. My respect for his knowledge-base and ingenuity is vast. Such a pity it's wasted on constructing props for his bizarre pet theories of weight control.

While the 7.7% of calories as fat in both studies is something which amuses me greatly, I do have to admit it may just be an hysterical accident.

At least I'm up front about my rather pronounced personal biases and rather peculiar sense of humour.



raphi said...

so the only low-fat diet than 'works' must be so low-fat that it's nutrient deficient. and would lead to increased rates of suicide in most adherents due palate-boredom.


valerie said...

The Hall 2015 study lasted only 6 days. On such a short horizon, I would expect the low-carb diet to lead to a lower fat loss simply because the body is busy burning glycogen at first. Only when glycogen is (relatively) depleted does the body move onto fat reserves.

On the low-fat diet, the glycogen stores at not depleted, so the energy deficit would come straight from the fat stores.

The real question (which they do not address in the paper) is which diet makes the body lose the most overall energy (glycogen + body fat).

Passthecream said...

Peter, Peter, (Wags finger like an Italian Nonna)

Anyway, you can throw away all your keto shakes now and mix this Carboshake(tm) instead:

For 1000 food cals, quick back of a research paper calculation gives

1 litre of ordinary CocaCola(tm)
2 small ie 50g eggs
45g gelatin

Mix well, cook to kill bugs if the liquid doesn't do that. Who says you can't have your coke and heat it?


Peter said...

Pass, I think I need that comment in my Christmas crackers this year!

Valerie, yep, the second three days out of six on a mild LC induction doesn't tell us much but I still want to know how they got what they got, is it possible to understand in any sort of detail?

raphi. yes, looks that way. But it does pose questions as to how a nutrient bereft diet can be both satiating enough to give massive weight loss......... Dives me back to caloric access to adipose issue and hypoisulinaemia. It the moment I'm deep in to first pass hepatic insulin extraction.


Peter said...

Ignore the "both" in the above please!

Passthecream said...


I do know that if I tried that I wouldn't survive very long. It is representative of my dark, dark years.

Looking back at Insulin makes you hungry (7) you proposed "Satiety occurs when the brain senses that calories are no longer being accepted by the peripheral tissues using an ROS signal. Superoxide will be that signal." CNS insulin not involved etc.

What's wrong with this rough scenario:

Very healthy metabolonormative individuals with responsive beta cells (assuming that's enough fat to empower them).
Low fat of any kind intake = high insulin sensitivity in the hungry cells.
Big intake of mixed carbs including gluc and suc -> short delay then big insulin release -> insulin driven glucose input with no lipolysis ( no exogenous lipids to burn either) but still enough atp and proton increase from the big pulses of gluc and insulin to give a big superoxide signal -> insulin receptor shutdown -> as above, high level of unused circulating nutrients etc etc.

I'm sure that there are many many devils in the details and it would have serious downstream consequences.

Passthecream said...
This comment has been removed by the author.
Passthecream said...
This comment has been removed by the author.
Passthecream said...

A little later: I re-read

Truly masterful explaining. Perhaps don't try to achieve carbosis whilst using metformin?

Chris said...

Body fat loss on the isocaloric diet was calculated as a difference between net fat oxidation and fat consumption, raising the question of what replaced the lost body fat to maintain weight. The methods here seem designed to ignore lipogenesis from carbohydrate.

The 30% CR diet results are particularly weird, though. It especially bothers me that the didn’t keep their methods/reporting entirely consistent with the isocaloric diets. Measured weight changes and protein balance and whatnot vs not measuring/reporting, and DXA vs metabolically calculated fat loss. And some of the results are a bit unexpected. E.g. carbohydrate burning increased in LC subjects over the course of the study after an initial decline, but with no change in protein balance in the same period. Might just be noisy data or whatever, though.


Weight loss in LF was predicted to be about 250 g over six days, or about half what a 30% CR diet suggests (fig. 3F), but actual weight loss was about five times the predicted value, or around 1.25 kg. This difference might be understandable in an LC diet because of glycogen loss, but that’s really weird in HCLF diets unless they’re just sort of dehydrated overall. Doubly so since DXA is biased toward higher FFM and lower FM in dehydrated tissue, per "Accuracy of DXA in estimating body composition changes in elite athletes using a four compartment model as the reference method", Nutrition & Metabolism 2010, 7:22.

Did they maybe accidentally give a bunch of pre-diabetic obese people mild polyuria with a low-fat, calorie-restricted diet?

Peter said...

Chris, when the Hall study first came out I sort of looked at it, thought well, the diets are chosen a bit oddly, it was only 3 days at the peak of Atkins Flu and bully for Hall if he managed to get numbers to support his twisted view of life. Let's move on and get some real science done. that more or less seemed reasonable as a conclusion.

I only got any more interest in the original study when he wrote to the BMJ bashing Ebbeling et al's work. Even that would be a matter of an "Oh well, what would you expect?" sort of a response. But the the core, number one "They wuz rong" point made by Hall is utterly incorrect. So we are now back in to the dichotomy of "Is he stupid?”, ie he really believes that Ebbeling should have use “Pre” as the anchor point or is he so biased that he knows his objection is utterly incorrect but he's going to make it anyway because enough people are likely to believe him anyway (Hall’s the one that showed LC doesn't work dontchano). This might be politely called bias.

If the rest of the study starts to fall apart under scrutiny of the detailed methods (way beyond me) it would push me towards the latter opinion. Ultimately I worry that Hall is very, very bright and would never make a mistake unless he thought any net benefit was in his favour.

But I may be wrong.


Passthecream said...

Peter, Slicing and dicing the dietary ingredients it certainly looks like the macro ratios in the Hall paper you reference were very carefully chosen to put the baseline diet within easy reach of that magical 7% on the HC arm while the LC version ends up not breaking through the Lutz barrier of 72g/day in fact ~twice as high. 


Should note though that in the results section it says "only the RC diet resulted in a 22.3% (+/- 7 p0.001) decrease in daily insulin secretion" and "only the RC diet led to significant sustained adaptations of carbohydrate and fat metabolism"

Aren't those the important results? And without even pushing the boundaries of serious LCHF! Who gives a rats' about the minor differences in weight loss? I want lower insulin please. And how hungry were those low fat eaters?

There is this older study by N. Hwalla et al which is similar but credible.

The Hall high fat (RC) has: 
Total Cals 1918 with C29% F50% P21%
Hwalla's high fat (HM): 
Total Cals 1835 C35% F45% P20%, slightly more carby.

Hall's HC:
Total Cals 1918 C72% F7% P21% 
Hwalla HC:
Total Cals 2003 C60% F20% and P20% , much more fatty!

Interestingly if you add the calories from the lost fat mass back to both branches of Hwalla et al you get nearly the same total calories in each HC 2863, HF 2839 

Both studies run  for the same 4 week period. Hwalla et al selected IR subjects with high BMI and at the end of the trial "Fasting insulin levels, insulin to glucose ratio, and Homeostatic Model & Assessment Index decreased to normal ranges and were significantly lower in the HM group as compared with the HC group." and "Waist circumference measurements showed a significantly higher decrease on the HM than HC diet"

There is a more recent Hall et al study in which the macro ratios actually were in Keto territory. 
In Table 1 there the (higher fat) diet has 5060g sodium versus 3060g in the baseline. Big diff. Most fats don't contain a lot of sodium so this must have been mixed in with the carbs??? This gets more interesting if you look up Hall et al's  modelling paper 

It is complex but the explanatory detail for equation 3 says  in part " I assumed that a  5000mg/day change in sodium intake would be balanced by a 1.7 litre change in ECE ...." (ECE is extra cellular fluid mass) and also "removing dietary carbohydrate doubles the sodium excretion rate compared to a low sodium diet"

Something going on there do you think?

Now please excuse me while I uncross my eyes and take my treadle powered lawn mower for a spin - I need the excercise.

Peter said...

Enjoy the lawn mowing. Hall is very, very clever.......


Passthecream said...

ECE - should have typed ECF.

Fats and sodium - could have been soap I suppose.

adamkern8 said...
This comment has been removed by the author.
Peter said...

adam, I think you could say that for people with metabolic syndrome. But are you think about a "Hall" type study? Don't forget, Hall's approach tells us nothing, nothing at all, about spontaneous weight gain in humans. You might be interested in the db/db mice who ate a similar diet to the one you mention