Saturday, May 16, 2020

Low fat vs low carb again (2)

For your enjoyment I have simplified this graph from

Hyperinsulinemia Drives Diet-Induced Obesity Independently of Brain Insulin Production

out of Jim Johnson's lab:

















down to this graph (at great effort) to show only the mice on standard chow:


















These are the insulin responses to an IP glucose tolerance test at a year of age in mice which have been fed on good quality chow all of their lives. The mice in the top curve are phenotypically normal in their insulin response to glucose, the mice in the lower curve have had three out of four of their insulin genes knocked out. They weigh the same.

We all know from the Surwit posts that the normal insulin exposure mice have an 11% decrease in median lifespan compared to low insulin exposed mice, Jim Johnson's lab again.

None of us is in a position to have our insulin genes partially silenced from before birth, but we do have a choice as to how much insulin we expose ourselves to, based on our dietary choices.

What we need to know is what the insulin response to a given meal might be if we were to try to imitate the partial insulin gene knockout mice. Very few studies have provided this sort of information but the current pre print from Hall et al does just this. Here is the graph















The red curve is from a group of people fed a single meal of a mildly ketogenic diet. With insulin peaking between 20 and 30micoU/ml this is quite similar to the value in mice with reduced insulin gene load, those pan out at around the 22microU/ml mark (don't you wish everyone just used picomoles all the time? Well, I do). Or you can eat low fat, plant based and choose to expose yourself to over 100microU/ml of insulin. Doing this you might still lose a little weight (another post, eventually), you might lose a little fat but you also might lose a few years of lifespan as the insulin drives ageing with its associated chronic diseases.

How many years? If the median lifespan for humans is around 70-80 years and we are talking about an 11% reduction that gives us a ballpark of just under a decade lost. As a thought experiment.

PBLF, plant based low fat. ABLC, animal based low carbohydrate.

There is no choice.

Peter

Total aside: I really hope that Hall keeps the title of this paper unchanged in the version which eventually gets published. It's a single sentence of prose which encapsulates what is wrong with nutrition research. It is absolutely, totally factually accurate, while being completely selective in its choice of factual content to give an absolutely misleading impression. As a declaration of bias it is unbeatable. I love it.

31 comments:

ELau said...

Hi Peter,

How do we account for the significantly worsened ApoB and sdLDL numbers in the animal keto group? You have suggested in the past that this is an indication of damage being done. Surely a bit extra PUFA in a animal-based keto diet can’t do that much damage compared to a plant-based high carb diet?

Gyan said...

Low fiber content of the low-carb diets may have some effect via the FIAF pathway.
Plenty of low-carb people however vouch for fiber and lay importance on eating sufficient fiber through vegetables. I had been influenced earlier by anti-fiber warnings of the author of The Fiber Menace--Konstantin Monastyrsky.

He emphasizes more on the mechanical problems associated with excess fiber in the guts. But I wonder--is pancreatitis really due to mechanical obstruction of the pancreatic duct?
Does fiber really absorb water and swell upto five times its volume?

I still would agree that people should not over-do on fibers. I wonder however is how the dietary saturated fats play role here. People always used to eat wheat with saturated fats. Do the saturated fats have an action regarding fiber and also on the gluten and WGA sensitivity?

Indeed, I have been meaning to ask for an update on the Ebringer hypothesis of the autoimmune disorders? It has been more than ten years.

ctviggen said...

Elau, I'm not sure I'd be concerned about those. Markers often don't tell you anything.

For instance, markers like that are highly transitory. I tested Dave Feldman's cholesterol drop method by fasting 4.5 days and getting ApoB tested. I got 116 mg/dL. I then ate as much fat (mainly, saturated fat via dairy; drank several cups of cream) and calories (keto) as I could for three days, and retested after about 12 hours fasting. Got 83 mg/dL. My ApoA1 went from 130 to 146; my LDL-p from 1170 to 1006 nmol/L; LDL from 127 to 106 mg/dL; HDL 30 to 54; trigs 135 to 68. Even my Lp(a) went from 367 nmol/L to 304 nmol/L. CRP from 1.7 to 1.2 mg/L (and a later "normal" test of 0.4).

So, I went from apparently deadly fasting to even more supposedly deadly high fat, high (relative) saturated fat...and my numbers all got better.

Under Dave Feldman's energy model, these values are transitory. Mine are all over the map, low carb/keto since 1/1/14.

So, I think 2 weeks of any diet is hard to gauge.

And, if you want probably the best test to gauge actual heart disease risk, get a coronary arterial calcification (CAC) scan done. Not perfect, but is better than markers. I got a CAC scan done last year, got zero score, even though I have some "scary" markers.

cavenewt said...

Re Gyan's comments on fiber—

When fiber is mentioned, am I the only one wondering if they mean soluble or insoluble? Because they're almost entirely different things, right? Insoluble fiber is often described as eating sawdust — it's simply mechanical roughage that goes straight through. Soluble fiber, on the other hand, is a prebiotic and seems to be a Good Thing. Eating vegetables gives you both kinds of fiber. Eating wheat bran is the sawdust.

Frunobulax said...

@Gyan, I always think that we're built to eat what our hunter-and-gatherers ate. If that's right, it's hard to imagine that some fiber would be bad for us. (The inuit are not representative for hunter-and-gatherer diets.)

Excessive amounts might be a different story (if you look at some keto bread, it seems to be 99% insoluble fiber, why would anyone eat such a monstrosity?). And it's possible that we do well on a zero-fiber diet. Most nutrients have a U-shaped optimality curve though...

Personally, I do eat some fiber. Protein seems to be fairly insulinogenic for me. (I went carnivore for a while, with maybe 35% energy from protein, and had some effects of a carby diet.) So do limit my protein and eat salad/vegetables with plenty of butter/oil, as this seems preferrable to drinking oil or eating butter :) Seem to work better. Hard to tell if the fiber in the vegetables has some effect.

Gyan said...

I suppose the longevity benefit obtains only for low-PUFA low-carb diet. Which the ABLC is not.
Or the low-PUFA low-carb diet might show even a greater benefit?

How would the insulin plot look like with low-PUFA instead of the high-PUFA ABLC?

Gyan said...

That a low-fat diet suppresses appetite leading to lower calorie intake is surprising.
Also, are the insulin values lower than the values on the normal diet? From the previous posts on carbosis, I gathered that a real low-fat diet increases insulin sensitivity resulting in lowered insulin production.

Why are the markedly lower insulin values in low-carb diet unable to suppress appetite? Because of high PUFA content? Would one see lower calorie intake in low-PUFA low-carb diet?

ELau said...

Hi ctviggen,

Thank you for the comments. I’m aware that lipid particle numbers alone may not cause CVD. But once artery gets damaged due to any number of reasons, how much of an effect ApoB particle numbers will have on risk will be anyone’s guess (I think). So imo it is worthwhile to monitor those numbers as one ages/go through lifestyle changes or medication changes.

Another point I wanted to mention is that from what I’ve read and also my own testing, multi-day fasting often results in a transient but significant deterioration of lipid numbers. However, fasting is an acute stress and whatever changes it may create probably shouldn’t be grouped with less aggressive modifications such as diet.

Considering all that, I would love to know if you’ve tried any n of 1 experiments that is similar to the aforementioned paper but using a longer time frame. Whether or not a major persistent change in diet causes a long term change in lipid numbers, and why, would be really interesting to explore.

Would also love to hear what @Peter thinks on this topic.

Elau

LeenaS said...

Hi ELau,

As far as science goes, I'm concerned on ApoA/ApoB ratios only. This ratio is the only marker in cholesterol data which has direct & dose dependent correlation to heart attacks (Search Lancet 2004 & INTERHEART by Yussuf & al). Reading it in time, thanks to Dr. Kendrick, I did take mine in 2004, after four years of lc eating, which then did have more plants than the present butterfat & animal fat based food that I have eaten for the last ten years.

According to labs, my numbers are getting only better and better - and they were quiter decent to start with. As a senior I'm quite happy to belong to the worlds lowest quintile of this ratio. Also, by comparing the ApoB and LDL numbers it is easy to see, that LDL particle size has grown quite nicely.

Cheers,
LeenaS

ELau said...

Hi LeenaS,

Thank you for the reply. Congrats on the improvements! But to get rid of potential confounders such as unknown factors acting on your system gradually over time... Was this diet-induced improvement a noticeable step change that persisted when you switched to animal-based from plant-based LC diet?

Elau

Peter said...

Hi ELau,

My view would be to ask what a diet high in PUFA might do to the linoleic acid content of both my chylomicrons and VLDL particles. In addition to stressing about what it might do long term to my inner mitochondrial membrane lipid double bond index. The low insulin is good, clearly, but the main problem with the lipid numbers is that they have been measured.

I am in the happy position of being utterly uninterested in lipid numbers. Getting your fat as linoleic acid is something I'd worry about long term, short term it would be HbA1c as a surrogate for insulin exposure...

Peter

Peter said...

Gyan,

There are several posts to come.

Peter

Peter said...

ELau,

Just on a matter of semantics: Fasting produces a "significant deterioration" of lipid numbers. The presumption here seems to be that fasting makes your lipids try and kill you, which is bollocks. So we are labelling a "change" on fasting with a value judgement. We thing we know the significance of the change. So clearly there is a paradox. The answer to which is that the lipid hypothesis, from beginning to end, is bollocks and your lipid changes on fasting are not a paradox. They are simply adaptation to survival. The paradox is why anyone believes the junk Keys spouted last century. Oh, and clearly if you have pre-loaded your adipocytes with soybean oil, you might be in trouble. But then, you would be whatever you do.

Peter

LeenaS said...

Hi ELau,
I was not on plant based LC to begin with, just on an LC with more plants and far less fat than now. My few years as nearly a vegetarian taught that lesson well enough decades ago. Plant based was my life always before lc, so that even the first move to more animal based food and lc worked wonders, in 2000. But no, there are no real confounding factors other than food between the two measurements mentioned. ...if you do not count aging as such; yet that should do exactly the opposite to what has happened to me :)

Cheers,
LeenaS

Frunobulax said...

@Gyan I don't think "appetite surpressing" is the correct term here. I assume they were hospitalized (not sure if it was mentioned), they are called 5 times a day "food is on the table", they will eat whether they are hungry or not.

It needs way more than 2 weeks until we start to change our eating habits. It took 6 months for me, during my first 6 months on keto I ate much, much more than I do now. Perhaps also because I found the food more satisfying than before. I still lost weight...

And we consume more calories on keto anyway, because we burn/need more energy, even if we eat only when we're hungry. Which is not a bad thing. Who gives a s how much calories we eat? Important is how weight and blood markers progress.

So both is wrong: (a) calorie counting at all and (b) measuring "appetite" via calories if you feed participants 5 times a day. Unfortunately, 2 wrongs do not make 1 right :)

raphi said...

Let me relay some comments Kevin Hall (KH) made on my twitter feed today, for what it's worth:


Alex Leaf criticized the study saying "Why would the blog author believe a liquid meal test is representative of anything practical? It's highly unlikely that such an insulin response would occur when eating whole foods." and "I definitely believe that some people would respond poorly to whole food sources of carbohydrate. My criticism was oriented towards the author's focus exclusively on the liquid meal test providing 30% calories"


KH then responded to Alex saying "I accept the criticism regarding the liquid mixed meal tests, but postprandial CGM measurements of interstitial glucose after 762 ad lib meals nicely corroborate the blood glucose findings from the mixed meal tests suggesting they may have been representative of the overall diets"

Eric said...

https://www.theguardian.com/society/2020/may/20/type-1-diabetics-type-2-coronavirus-nhs-study

Can we understand this from a protons point of view?

Frunobulax said...

@Eric I mentioned before that the death rate from COVID-19 seems to correlate well with the coronary calcification (CAC scans). As it turns out, T1D seem to develop more calcification than T2d. https://care.diabetesjournals.org/content/26/10/2923

I admit that this surprises me too. After all, we recommed eating carbs to people that have no natural ability to adapt to glucose levels in the blood. Any man-made insulin therapy is got to be worse than what our body can do, even with a functioning pancreas.

So this would indicate that hyperinsulinemia is less significant a factor for calcification than bad glucose control? But I don't know enough about T1D treatment. Maybe we give those poor people too much insulin as well as too much glucose?

Peter said...

Hi Eric, in type 2 there is insulin insufficiency due to the pancreas' inability to overcome insulin resistance. In type 1 there is no limit to the insulin available in the bottle, you just need a bigger syringe. Not Protons, just practicalities and acceptance that CVD is diabetes in-situ, which it is.

Peter

ELau said...

Hi LeenaS,

Cool paper... Lots of learn...

Peter,

Thanks for pointing out my bias!

ELau said...

@LeenaS,

Hi, I actually looked into some of the details of The INTERHEART study. While ApoB/ApoA1 ratio has a good association with MI by OR, it would also seem that diet factors such as FRUIT INTAKE (horror) has a close-to but inverse OR curve. In that more fruit consumption is associated with reduced MI incidence. It would indicate that either eating more fruit is almost as good for you as having a low ApoB/ApoA1, or this study isn't powered enough to be super helpful.

LeenaS said...

ELau,
have another look at the INTERHEART odd ratios, and also note that "the fruit & vegetables" actually meant either none or some daily. Fruit consumption lies far, far behind the Apo ratios. Yussuf &al actually confirmed that fruit/vegs was only a minor modifier in the newer PURE study.

Cheers,
LeenaS

ELau said...

Hi LeenaS,

I did take another look at INTERHEART, the specific paper is Dietary Patterns and the Risk of Acute Myocardial Infarction in 52 Countries. In it, fruit intake is in 4 quartiles and adjusted OR for highest intake is about .7. Vegetable intake in the highest quartile is also 0.7 for adjusted OR, which do make them close to the strengths of ApoB/ApoA1 ratio associations. I dug around some more and there are more details in regional papers that came out of INTERHEART, eg. China, where the fruit intake quantity has been specified. More than 7 times per week brings adjusted OR (95%CI) down to 0.4.

The PURE study: I haven't found a paper that talks about ApoB/ApoA1 other than one that associates the ratio with diet (not super strong associations). So there isn't an apple-to-apple comparison in terms of strength of associations between ApoB/ApoA1 and diet...

LeenaS said...

Hi ELau

In INTERHEART, the OR of the Apo ratios between the lowest and the highest quintile is more than 4. This is quite a bit more than 0.7 going the other way : )

In a lecture after publication of the PURE study, now taken away from free internet, but well summarized by Dr Eades in his blog, Dr Yussuf himself said that according to their PURE results, one fruit a day is possibly good, and if you like fruit, why not take another. More than that yields no more benefits.

Apo ratios are an easy test, they are related with a linear and dose dependent risk correlation (unlike any other cholesterol lab test), and they seem not to depend on age, race or sex, which makes comparisons very easy. Maybe this is why they are not recommended by medics?

Cheers,
LeenaS

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

Hi LeenaS,

The ApoB/ApoA1 OR of 4.0 is for the 10th decile compared to the 1st decile, a huge 2.5 SD difference in the population whereas if you look at a smaller SD difference (eg. 1 SD) you will see that the OR is closer to fruit intake in the China study, the data in which provides a more detailed and comparable 1 SD. (pls see: Influence of dietary patterns on the risk of acute myocardial infarction in China population: the INTERHEART China study).

LeenaS said...

Hi ELau

Sorry, I'm more interested in places where I know something about the consumption and culture. Food consumption in China is very different, so I find it quite hard to make comparisons to our ways.

Ho much fruit they did consume in the highest quintile in China?
According to Dr Yussuf, fruits are common only among the populations with high standard of living.

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

Hi LeenaS,

I looked at the China paper because it provided more color to the main paper which did not break fruit intake down. The highest quartile intake in the China group is more than 7 times a week. (Quartile 4 intake folks is a small %of both AMI and control groups). The adjusted OR by quartile is 1, 0.7, 0.6, 0.4.

Going back to the main point, even if ApoB/ApoA1 has a stronger association, it is not dramatically stronger. And it seems to me that it may be easier to invert the causal relationship in fruit intake than it is in ApoB/ApoA1, as in it’s improbable that heart healthy people tends to eat more fruit. On the other hand, There obviously could be bigger confounders relating to fruit compared to ApoB/ApoA1, eg. As you mentioned socioeconomic being one that would be hard to adjust fully for.

Looking forward to your thoughts...

LeenaS said...

ELau,

So, in the highest quintile in China they ate seven fruits or vegetables - in one week :) Is that not what nearly everyone does here? And, it does not get any better after that. More fruit brings no further advantage, says Yussuf. One piece a day...

Apo ratios are much more workable with food as a whole (and not just the one 100 calories from one piece of fruit or vegetable a day). Very high milk- and animal fat has worked wonders, we've found. Furthermore, you can stop at looking other cholesterol markers, which really do not tell that much.

ELau said...

Hi LeenaS,

Thank you for the back and forth discussion. Just to clarify the highest quintile eats seven fruits a week, that's not very prevalent as it is a proportionally much smaller segment of the groups...