Insulin resistance (16) Yes. Vitamin E can cause weight loss

Part 3

It's time to set out a logical explanation for this graph:

taken from 

Ability of high fat diet to induce liver pathology correlates with the level of linoleic acid and Vitamin E in the diet

The first thing we have to do is to ask the correct question:

To me the correct question which has to be answered is why an high fat diet (35% of calories) with only 1% of calories from LA is obesogenic, at all. If you come from the point of view that the mechanism of obesity is down to LA alone, irrespective of macros, you need to be teasing this paper apart in great detail.

We know from

Differential effects of saturated versus unsaturated dietary fatty acids on weight gain and myocellular lipid profiles in mice

that a mouse eating an high fat diet based on D14521, if you choose your fat correctly, will not become obese at all provided LA is limited to 1.4% of energy intake. We can simplify Figure 1 from the paper:

to this:

This makes it abundantly clear that if you hold one variable constant, that is the percentage of energy supplied by linoleic acid, the nature of the rest of the fat in the diet is not insignificant.

Okay.

Here is the fat used in Graham's study, which is obesogenic with fat at 35% of calories:

and here is the non obesogenic diet with 45% of calories from fat in the D12451 study. I've combined the saturated fats together and expressed all as percent of total energy to allow easy comparison with the obesogenic diet from Graham's study:

They are identical in composition, the only difference is between 35% and 45% of energy from fat causing an absolute difference in overall calories from fat. A complete coincidence, but still very neat. Both studies look like they could have used the same jar of fat to make up the two diets. Except...

Those 27% of calories from saturated fat in Graham's study were largely from hydrogenated coconut oil. In total, hydrogenated coconut oil supplied 89.1g/kg of saturated fat to the diet.

We can ignore all fatty acids of C16 or greater as they are, as always, completely absorbed as chylomicrons through the thoracic duct and in to the systemic circulation, bypassing the liver completely.

For the shorter chain fatty acids I think we can ignore the myristic acid (~17% of the coconut oil) as it too is mostly carried in chylomicrons and treated much like palmitic acid. Only around 5% of it goes directly to the liver as FFAs.

The lauric acid which makes around 45% of coconut oil is partly packaged in to chylomicrons but something around 30% is delivered directly to the liver as FFAs.

Capric acid (C10) and caprylic (C8) (~15% combined in coconut oil) are completely transported directly to the liver as FFAs and do not enter the systemic circulation.

This medium chain triglyceride inclusion is the major difference between the 35% fat obesogenic diet used by Graham et al and the non obesogenic diet based on the stearate/palmitate/oleate mix in the cocoa butter D12451 study.

This matters.

To get some idea of what is happening we have to go somewhat in to reductio as absurdum and use pure MCT oil with a splash of soybean oil giving 50% of calories as fat in this study

A rich medium-chain triacylglycerol diet benefits adiposity but has adverse effects on the markers of hepatic lipogenesis and beta-oxidation

and look at this, Figure 1:

from which we can extract the relevant lines like this:

Here we can see that, on 4% of linoleic acid, MCT oil at 45% (no lard in this group) of energy intake is obesogenic when compared to 5% LA on a low fat background ie without the MCTs.

MCTs, in any significant amount, are obesogenic in their own right.

In this last study the obesity is made much worse by adding lard containing significant LA, that's what the faded-out lines on the unaltered graph show. I'll just add in the all-lard fed group as we need it when we look at pAKT levels:

So now we can look at the degree of insulin signalling present in hepatocytes at the time of euthanasia after a six hour fast. The full panel C from Figure two looks like this:

which we can simplify down to the interesting bits like this:

Clearly there is a marked reduction of insulin signalling in the lard fed mice. They are the most obese and are providing long chain fatty acids to the fasting liver, some LA but also palmitate and oleate. This is the normal physiological insulin resistance of fasting augmented by elevated basal lipolysis.

The lard-free highest MCT oil fed mice also demonstrate hepatic insulin resistance but far less than the lard fed mice. Which looks like a paradox.

The main problem with interpreting the pAKT signal is timing. The levels were measured in liver tissue after a six hour fast, which is quite a long time for a mouse. If you feed an MCT rich meal you would expect to flood hepatocytes with FFAs via the portal vein during the peak absorptive phase, probably around 1-2h post intake.

It is this flood of MCTs, which enter hepatocyte mitochondria with minimal restriction, which necessitates the resistance to insulin. Never forget that insulin resistance, at it's core, protects against the damaging levels of ROS generated by unrestrained elevation of delta psi. Insulin resistance automatically reduces insulin catabolism.

MCTs are not stored, so the excess lipid within the liver or exogenous lipid from the systemic circulation by six hours post intake will reflect the highly controlled oxidation of longer chain FFAs, not the un-restrained catabolism of MCTs at peak nutrient absorption. As the mice consuming MCTs are less obese than those on lard, it is reasonable for the level of insulin resistance to be lower because basal lipolysis is also lower.

In fact there are various influences on the high MCT fed hepatocytes pushing insulin sensitivity in conflicting directions, especially under fasting. But we can tease out what matters, and when, from a few extra studies. 

The Protons prediction from hepatic insulin resistance with associated reduced hepatic insulin extraction is the facilitation of elevated systemic plasma insulin, primarily at the time of peak delivery of MCTs to the liver via the portal vein circulation.

We can get an idea of whether this genuinely happens from this human intervention study here:

Thermic effect of medium-chain and long-chain triglycerides in man

which gives us this graph:

There is a small but significant rise in systemic insulin after MCT ingestion. It was triggered by 400kcal of MCT oil. You can enhance the effect in rats, who can't object to the GI distress caused, by giving them close to half a full day's calorie intake as a gavage. That's in this paper

Relation of ketosis to metabolic changes induced by acute medium-chain triglyceride feeding in rats

including this effect on systemic insulin

which we can tidy up like this:

It's also quite simple to ask how much insulin is extracted by the liver from the portal vein before being passed out through the hepatic vein and in to the general circulation. All you have to do is compare systemic insulin levels after an MCT bolus in a normal person with those generated by people with severe cirrhosis and multiple porto-systemic shunts, which by-pass whatever dysfunctional liver tissue they have remaining in-situ. That will be this study:

Stimulation of Insulin Secretion in Manby Medium Chain Triglycerides

After just 30ml (270kcal) of MCT oil we get this doubling in systemic insulin in cirrhotic patients during the time of peak MCT absorption, but the effect is gone by three hours post ingestion:

This effect is only present under MCT ingestion. Neither corn oil nor other LCFA containing fat sources do the same. My expectation is that, at peak MCT absorption times, flooding hepatocytes with MCTs will generate an high ROS signal, confined to those hepatocytes. This ROS generation can be dealt with by storing the acetyl-CoA as intra-hepatocyte LCFAs derived from acetyl-CoA (to be later exported as VLDLs), by off-loading acetyl-CoA indirectly as ketone bodies and, finally, by simply resisting insulin.

Which allows systemic hyperinsulinaemia.

The action of insulin is the inhibition of lipolysis.

Which is obesogenic.

The Surwit diet does this.

Here we have a fundamentally different form of obesity compared to linoleic acid induced obesity. In LA obesity the fundamental problem is at adipocyte level and here *inadequate* FADH2 driven ROS generation allows excess insulin signalling to distend those adipocytes.

In Surwit's hydrogenated coconut oil diet derived obesity, the fundamental problem is an *excess* of ROS from the MCTs in coconut oil, delivered in high levels to the liver only, allowing passage of insulin through the liver to give systemic hyperinsulinaemia which acts directly to cause simple obesity.

I suppose that the addition to this is that the lauric acid which reaches the systemic circulation might be a factor in the obesity, the shorter chain length gives poorer ROS generation compared to palmitate or stearate. 

MCT obesity is the only form of obesity which should be *reduced* by limiting ROS signalling.

Drop the ROS, restore hepatic insulin sensitivity, allow the liver to extract insulin, so shrink peripheral adipocytes via normalisation of peripheral insulin levels. *Hepatic* ROS scavenging does this:

and the converse is ROS scavenging does this:

Or, more relevantly

Or, with a minor change of one arrow's colour:

I'll leave it at that for the time being. There are, as far as I can find, no studies looking at the effect of vitamin E on Surwit-like diets. That's understandable. Why should vitamin E, considered to stabilise PUFA, have any effect on saturated fat induced obesity? So why bother looking at this?

But, given the ROS hypothesis of obesity, a plausible mechanism for the action of Surwit like diets is clear.

Peter

Insulin resistance (15) Vitamin E for weight loss? No

Part 2

Here is the next study.

α‐Tocopherol suppresses hepatic steatosis by increasing CPT‐1 expression in a mouse model of diet‐induced nonalcoholic fatty liver disease

From the methods:

"Experiment 1: Mice were divided into seven groups (n = 10 in each) and given the following diets for 8 weeks: standard diet (control group; 30% protein, 68% carbohydrate, and 12% fat including vitamin E acetate [500 IU/g]; Research Diet); HF diet (HF group; 20% protein, 20% carbohydrate, and 60% fat including vitamin E acetate [500 IU/g]; Research Diet) and HF diet with α‐tocopherol (α‐Toc) which is one of the natural vitamin E forms supplementation (20, 50, 100, 150, and 200 mg/kg)."

Okay, a ghastly typo.

We have no idea which of the Research Diets these mice were fed on. I am going to assume that the chow resembles PicoLab Rodent Diet 20/LabDiet 5053 and contains, as per my last post, ~100iu/kg of synthetic vitamin E acetate yielding, also as per last post, 44.5mg/kg of active vitamin E in the food.

The typo is to state that this chow contains 500iu/g. That's quite a lot of vitamin E. The correct amount is (almost certainly) 100iu/kg, not 500,000iu/kg.

I feel it is reasonable to assume the high fat diet was something similar to, or in fact was, D12942 which also probably contains around 100iu/kg vitamin E. Obviously the mice would eat less weight of D12942 than chow because they eat to caloric need. This will be met by a lower weight of D12942 so their intake of d-α-tocopherol would also be a little lower than if they ate the chow. More like 40mg/kg if you taken in to account the reduced weight of food eaten.

I am also going to assume that they added their supplementary vitamin E to this standard high fat diet so we're looking at intakes based on diets containing totals of 60mg/kg to 240mg/kg of d-α

-tocopherol per kilogram.

Anyhoo. For the time being I'm going to ignore the changes in everything other than total body weight.

There are no data presented for the effects of most of the supplement levels used, though these were recorded as per the methods. I think it's safe to assume that the effect on weight was consistent across all vitamin E intakes used, otherwise they would have mentioned it. This is what they actually presented:

There are, undoubtedly, effects from vitamin E supplementation on parameters other than total body weight. People may find the liver damage induced by high dose vitamin E fascinating. I do. But that's another story.

So I think we can say that, in a poorly described study, vitamin E supplementation has absolutely no effect on the body weight of mice over eight weeks of feeding an high fat diet. Over a wide range of dose rates.

To continue the catalogue of appalling vitamin E focused studies, it's now time to look at this one:

Effects of d-α-tocopherol supplements on lipid metabolism in a high-fat diet-fed animal model

How bad is it? It's this bad:

"After the adaptation period, the mice were randomly divided into three groups. Nine mice were placed in the control group [CON, regular diet (10% of calories derived from corn oil) and distilled water as a vehicle (0.1 ml, p.o.)]. Another set of nine mice were placed in the high-fat group [HF, high-fat diet (45% of calories derived from lard) with distilled water as a vehicle (0.1 ml, p.o.)], while the rest of the mice were placed in the high-fat diet with daily oral administration of 100 IU/kg B.W. of d-α-tocopherol group [HF-E, high-fat diet (45% of calories derived from lard)]."

So we know nothing about anything. We have no idea of the vitamin E levels of the control chow or of the high fat diet. We don't even know if the high fat diet was manufactured specifically or whether they just added lard to chow to make 45% of calories from the lard which diluted the chow's vitamin E. We don't know what the lard was composed of in terms of LA either. Or even if it was Japanese or from the USA. We *do* know it was high enough in LA to make the mice fat.

None of this matters too much because the vitamin E supplementation was given by oral gavage of 100iu/kg once daily. This was pure d-α-tocopherol so the arithmetic is easy. The 100iu gavage provided 67mg of active d-α-tocopherol, not the racaemic mix and not the acetate ester. The mice weighed 32g so each got ~2mg/d by the end of the study.

If we reverse engineer to translate this in to how much vitamin E would need to be added to food to deliver that same dose we can do this. We can say that a mouse eats ~2.8g/d of high fat diet. So there would need to have been 2mg in 2g of food or 1000mg/kg of food. Though if you used the synthetic acetate ester then around twice that. This is a massive dose of vitamin E and guess what effect it had on body weight? You're waaay ahead of me:

We might also take note of the caloric intake per day which comes out as exactly what anyone would expect for mice on an high fat diet, in contrast to the last post. The current study calorie intake is picked out in blue:

I hope you're not getting too bored with this. I suffered for weeks with these studies. Now it's your turn. 

I guess I'm not selling you vitamin E as a weight loss hack. That's good.

So how do we square these studies (and many others, it's been a rough three weeks of reading) with the results from

Ability of high fat diet to induce liver pathology correlates with the level of linoleic acid and Vitamin E in the diet

where there is a marked decrease in weight gain on an high fat diet with modest vitamin E supplementation?

I absolutely accept that these data are correct as reported.

These are the sorts of findings which test your hypothesis of obesity. It's what makes slogging through the typos and brain farts and shifting definitions of high vs low vs unspecified levels of vitamin E in diet trials worthwhile.

Ultimately we are looking for circumstances where reducing ROS with vitamin E allows weight loss. Some weight loss anyway.

I think there might be an explanation.

Peter

Insulin resistance (14) The beginnings of vitamin E for weight loss

Part 1

I've spent an inordinate amount of time on this, and related, studies over the last few weeks

Ability of high fat diet to induce liver pathology correlates with the level of linoleic acid and Vitamin E in the diet

There are serious problems with panel A of Figure 1:

Ordinary mice on chow typically eat in the region of 14kcal/d, ie 98kcal/wk. High growth rate mice can eat 18kcal/d, ie 140kcal/wk. So the chow column is plausible as energy intake per mouse per week. Some mice are reported to eat as little as 9kcal/d on chow. There's a range.

What is not plausible is that mice on high fat diets ate 20% less total calories than the chow fed mice while gaining 20-40% more bodyweight, all as adipose tissue. This doesn't make sense.

In general rodents on obesogenic high fat diets consume very slightly *more* calories per day over the weeks than chow fed mice. The weight of food needed to get those calories will obviously be reduced because there are extra calories per gram. So calorie intake usually ends up slightly increased but on a lower food ingestion weight.

Just for illustration here are the energy intakes from mice fed a 45% lard based diet discussed in the next post. They are typical. Outlined in blue are the numbers we need:

Call me a Cicotard if you like, but the panel A graph from Figure 1 suggests either massive, and I mean massive, uncoupling in the chow fed mice. Which I doubt. Or a catastrophically reduced daily energy expenditure in the high fat diet fed mice, *irrespective* of linoleic acid intake. Which I also doubt. Greatly.

It almost looks like there has been a mix up of weight of food consumed and the total calories consumed per week.

But that doesn't reverse engineer easily. If we say the 110kcal/week specified for the chow mice is actually 110 grams/week then that's 16g of chow each day. No mouse can eat this much chow per day. If we assume that it's for the whole of the group of four in the cage we get a food intake of 4g/d/mouse. Which is plausible.

But who knows? That's a lot of guesswork.

It's a bugger because we need to know how much food each mouse ate in order to determine how much actual vitamin E each consumed.

We are given the vitamin E per kg diet, and the number looks plausible. Looking at Table S2 we have the data for the chow (PicoLab Rodent Diet 20/LabDiet 5053) and can calculate the value for this well recognised and standardised diet.

Diet 5053 contains 99iu/kg of vitamin E acetate on the product sheet. This gives, assuming synthetic vitamin E, 89.1mg/kg vit E. From this calculator we have:

Also assuming the synthetic vit E is a 50:50 mix of the d and l isomers, we have, of active form, half that, giving 45.55mg/kg, as here in Table S2

On this basis I think it is very likely that the absolute amount of vitamin E in all of the diets in this paper are correct in Table S2.

If we deconstruct the (excellent) Table S2 to remove the arithmetic lines we can summarise the amounts of vitamin E in different diets like this

If we take Lab Diet 5053 to represent some semblance of a normal to generous level of vitamin E for normal growth and reproduction of rodents eating one of the most widely available chows in the world we have the low LA diets with less absolute vitamin E than 5053 and the high LA diet containing rather more vitamin E than the 5053 chow diet.

Working through the rest of the (largely appalling) vitamin E-HFD literature most groups would consider vitamin E at 50mg/kg of diet to be low dose and 200mg/kg to be high dose. Those would be dose rates of d-α-tocopherol. So we are looking at low versus moderate in this study.

Now we are in a position to look at the weight loss effects of various levels of vitamin E in various diets.

Peter