Tuesday, March 10, 2009

Cholesterol within nations studies

These are the slides from the within-countries discussion on cholesterol and heart disease. I've allowed the sarcasm back in, which was strictly limited when the slides were originally presented. OK, there is a correlation. In fact, if you are a bloke, having a cholesterol above that certain magic number on the graph is clearly catastrophic and boy, are you in trouble. No statins to save your life in those days!

But what is the mystery number which sentences you to cardiac death? Obviously the original Framingham study did not use a random scale. But the scale used in the study was highly non linear. Here is the same graph with the real numbers added.

Once you have realised quite how unlinear the scale is I would just like to draw your attention to that number on the bottom right written in RED. No, it's not a typo. That really is 1124mg/dl. Okaaaaaaaaay. Hands up if you think a TC of 1124mg/dl is part of a normal distribution of cholesterol values in a "normal" population. Are there any conditions which elevate cholesterol and increase the risk of heart disease? Answers on a postcard to any endocrinology department you happen to know the address of.

Anyway, next thing to do is to linearise the scale. I've been kind and left the upper data point in the middle, not placed it at the upper end, of the range quoted:

Now here is my main cheat, and I admit it's a cheat. There must be a population towards the upper end of that last data point who have medical problems to give them an increased risk of heart disease. Cushings Syndrome and hypothyroidism are two for starters. So I would argue that it is only fair to the representatives of this subpopluation that the risk scale is extended up for their benefit. 100 "events" per 1000 in this group seems possible, hence the extended scale:

So let's go back and look at the initial random points and leave the top cholesterol data point where it belongs, about a yard to the right of your screen:

Ah, that's better. Some semblance of honesty now. But again, not quite as convincing as the first graph. BTW did anyone notice that the left hand scale was events, not deaths?

Did the Framingham investiators look at deaths? Hahahahahahahahahahahaha bonk. Sorry, that's me laughing my head off. Of course they did. There is no association between elevated cholesterol and increased cardiac deaths, but the trend is that high cholesterol is protective. Luckily for the Framingham researchers they were underpowered to detect this. A whiff of the low powered studies we see nowadays.

Now let's just look at the MRFIT screenees. These are the many, many people who were looked through before cherry picking the victims for the MRFIT intervention trial which, incidentally, killed more people in the intervention group than died in the the usual care group. Luckily (again) this did not reach statistical significance, though it may have been of some biological significance to those extra people who ended up dead in the intervention group.

The original study had non linear group sizes (like Framingham) to specifically, oh, I mean accidentally, obscure the effects on all cause mortality in the low cholesterol individuals. The graphs above are taken from a later sub-analysis by a rather more honest and objective lead researcher. Such people do exist in cardiology.

The other hysterical aspect of these graphs is that the original data presented from the MRFIT screenees wasn't, wait for it......

It wasn't corrected for smoking!

Yet another re-analysis shows a marked association between TC and cardiac deaths, but only in smokers. About what you would expect if LDL cholesterol was doing its best to repair smoking induced damage and failing.

The association is still present in non smokers and is statistically significant but so much weaker that the biological significance is highly debatable. It probably represents sucrose intake.

So in summary, the original MRFIT screenees study (presented as the clincher for cholesterol causing heart disease) obscured the scary aspects of low TC and, err, forgot to correct for smoking!

Don't you love the foundations of modern cardiology! Can't sum up MRFIT better than Dr Werko.

For a breath of fresh air it's worth going to Norfolk in the UK.

This innocent little graph is plotted from the EPIC Norfolk data. Now I'd hate to suggest that being hyperglycaemic has anything to do with heart disease, but you can read the graph as well as I can. Association not causation. And of course we know that eating fat causes hyperglycaemia, just ask any diabetologist. Shrug.

The weird thing about UK researchers is that they give you the raw data, it's not a matter of a little table of regression coefficients corrected for this, that and the other. You can read the results tables and plot the graphs. They even give you raw smoking data. And of course you get both TC and that evil killer, the LDL level. Let's plot them on the same graph as the HbA1c vs relative risk of cardiac events.

Personally I'd be looking to minimise my HbA1c rather than my LDL-C.

But then I would say that.



Charles R. said...

Holy crap. That last chart is the killer--some pun intended. Yah, I think HbA1c might be something to pay attention to, huh?

Great analysis, Peter. And the sarcasm is perfect, and deserved. If I didn't know better, I would think that someone was attempting to slant the presentation of the data in order to promote a particular point-of-view.

But that would be cynical, of course.

Peter said...

The HbA1c or FBG data are very consistent across the board, whenever anyone has looked at it. Even in MRFIT for what that's worth.

Yudkin was right....


arnoud said...


Totally incredible how data is interpreted and manipulated to create a completely false picture of the relationship between cholesterol and heart cardiovascular heart disease!

Someone ought to go to jail for this (hint: not Peter).

I cannot bear to think of all the folks who were/are being sent down the wrong path.... clearly a very bloody battlefield.


gallier2 said...


The sad part, if you tell someone about this kind of find, you will be looked at as if you told you were abducted by aliens. People do not believe how incredibly misguided modern medicine is.

On the Framingham study, Dr.Eades wrote two excellent articles on it in 2006. If you hadn't read them here are the links


LeenaS said...

Thanks for graphs, peter!

"The sad part, if you tell someone about this kind of find, you will be looked at as if you told you were abducted by aliens. People do not believe how incredibly misguided modern medicine is.".. describes well the attitudes encountered, when I put a well documented Harvard study (NEJM 1997, on nurses heart incidences and fats) into graphs. It is so hard to believe that even the most respectable sources can do this. And even harder that they will do... Sad

Regards, LeenaS

ItsTheWooo said...

Oh wow, that last graph is truly shocking. Great post.
I have tended to think that cholesterol does have some pathological/facilitory role in CVDs but this has helped convince me blood sugar/insulin is pretty much entirely responsible.

Anonymous said...

I read recently that hunter gatherers in general have very low cholesterol levels. Total cholesterol ranges between 70 and 140! Is this evidence that lower is better? Does anyone have an explanation for why HG cholesterol is so low?

Peter said...

Hi Tod,

My wife has TC in this range, on the same diet as myself, but she is quite young compared to me and has now been LC eating for a significant portion of her life.

It's quite interesting that her immunity to infection does not seem to be as good as mine, also a feature of HGs???? You could just say I've been around a lot longer and have already had two sets of school bugs from my older children!

Don't forget that lipoproteins, especially LDL, are part of the innate immune system.

And of course the Kitavans, as subsistence agriculturalists, have lipids comparable to the Swedish population but without the heart disease.

So I think this all comes back to your lipids are what they are and anyone who thinks taking a statin get themselves in to that desirable HG level may be somewhat disappointed, especially if they are one of the unfortunates who contribute the the increased all cause mortality effect seen with low TC in Western populations!

As for concern re my wife's low TC, well I have just a little but then no one has ever looked at the effect of low TC in a Westernised human eating real food in a LC scenario and presenting with chronic normoglycaemia. I suspect under these circumstances lipids are irrelevant.


. said...

Dear Petro, you may be interested in these 2 documents with several statistics:



Chris Masterjohn said...


Interesting post. You so show how simply one can change the presentation of data to give a very different picture.

Your graphs do not tell the whole store, though. They group everyone with a TC level between 84 and 204 into one data point, when the Framingham study had zero risk below 150.

I think the totality of the evidence pretty clearly shows that oxidation of LDL membrane PUFAs is the primary and singly essential cause of atherosclerosis, although the causes of other forms of arteriosclerosis, plaque instability and rupture, minor alternative causes of infarction, and factors that can aggravate the effect of oxidized LDL are much more diverse.

(Summarized here: http://www.cholesterol-and-health.com/Does-Cholesterol-Cause-Heart-Disease-Myth.html )

Glycated hemoglobin would be a marker for high blood sugar, which is directly involved in oxidative modification of LDL, and the increased oxidative stress that leads to poor blood sugar control, which would also be involved in oxidative modification of LDL.

Pooling all the data together shows that total cholesterol is the poorest predictor and total-to-HDL cholesterol is the best predictor (out of the various cholesterol level markers), but of course the increments are fairly small and these are just markers; oxidation of LDL is the true culprit.

It is worth noting, however, that LDL receptor activity seems to be the principal governor of LDL oxidation. If the LDL isn't cleared from the blood, it will run out of antioxidants. Genetic defects in LDL receptor can lead to heart attacks in toddlers, children, and adolescents, while defects in the enzyme that degrades the LDL receptor reduce risk by 88%, almost abolishing it.


Peter said...

Hi Chris,

The zero risk with a TC below 150mg/dl in the Framingham study would be nice if we all lived in Framingham, but heart attacks with this level of TC do not seem to be unheard of in places other than Framingham, so I'm not so sure about that as a real effect...

I would absolutely agree that the oxidation of PUFA in LDL membranes produces multiple derivatives, especially those of linoleic acid, which are biologically active themselves and also signal the ingestion of LDL particles in to macrophages to produce foam cells. But here we are drifting away from cholesterol and in to PUFA toxicity. Keys was measuring TC, as were the Framingham investigators... They weren't asking about oxidation products of linoleic acid.

I would absolutely agree, if anyone suggests it, that glycated Hb is a surrogate marker for glycated ApoB100. But even within the Western diet, as consumed by Norfolk EPIC participants, that surrogate associates far better with not only CVD but also all cause mortality. If you are not glycating Hb you are not glycating ApoB100 and so not oxidising omega 6 PUFA associated with that ApoB100 shrouded LDL particle. I think it is also worth pointing out that Hb ids not intended to have its function specifically altered by glycation. The glycation of the glycation hotspot on the ApoB100 molecule inhibits its interaction with the LDL receptor, oxidises the particle's lipids and extends its plasma residence time. It probably gaurentees it ends up in a macrophage.

The glycation comes first. Even with the catastrophic level of linoleic acid overdose in the Norfolk EPIC diet, it is the glycation which appears to matter.

This leaves normoglycaemia at the core of any CVD reducing protocol. We know that low carbohydrate diets reduce inflammatory markers and that NFkappaB, a major control switch for inflammation, is activated in the liver by hyperglycaemia. We know that a whole raft of inflammatory mediators can be used to differentiate patients with advanced arteriosclerosis from those without it, and that the level of LDL cholesterol simply cannot do this. Possibly the level of oxLDL would differentiate the patients, but then you would be just looking at another product of glycation and the action of upregulated macrophages.

MTTP gene-493T polymorphism, which results in reduced levels of LDL cholesterol being secreted by the liver, results in increased risk of CVD. Researchers want to know why. The over expression of the LDL receptor which you mention decreases the risk of heart disease, medicine "knows" why and no one is looking for other explanations. They will be there.

This brings me to a very core concept. What is wrong with people with homozygous FH? Answer is that they cannot endocytose LDL particles. This is their genetic flaw. Consider infantile seizures from lack of neuronal glucose transporters. These people cannot transport glucose in to their brain and the problem can only be sidestepped using ketone bodies. Endothelial cells are not getting anything from LDL particles in homozygous FH. The question I ask myself is whether LDL endocytosis allows uptake of vitamins A, D and E and any of a few other molecules that endothelia cells might not be that well set up to make for themselves. Q10, lutein and xeazanthin come to mind, no doubt there are others. The fixation with the level of LDL in the blood has obscured the potential consequences of non functional LDL receptors on endothelial cells.

If we add to that the complication of clotting abnormalities being co-inherited along side dysfunctional LDL receptors in a significant number of FH patients, we end up right back in Oslo wondering whether the children with FH who develop CVD as infants have clotting abnormalities which get ignored in the excitement over easily measured TC values. I doubt anyone is looking. Incidentally the use of liver transplantation for FH treatment will provide a whole new set of clotting genes in the main organ which produces clotting factors. Plasma apheresis requires marked manipulation of the clotting system too, but I doubt anyone has looked at what they are doing beyond sieving the blood here.

I think the most important statement on FH has to be an early one from Brown and Goldstein, from their 1983 publication in Annual Review of Biochemistry:

"Among patients with familial hypercholesterolemia (both heterozygous and
homozygous) there is considerably variation in the rate of progression of
atherosclerosis, despite uniformly elevated LDL levels"

I have to say that I don't see LDL receptor activity as the governor of LDL oxidation. Sucrose and refines starches, with their attendant hyperglycaemia, appear to be much better candidates. That would fit with the EPIC data. High saturated fat diets appear to be associated with down regulation of the LDL receptor yet, within a given calorie intake, higher saturated fat intake causes least progression of arteriosclerosis, as in the American Paradox.

And of course the lipid ratios of the Kitavans, a population without discernable CVD, are not good. HDL is 0.5-1.1mmol/l in males and TC is 3.0-7.0mmol/l. A number of these individuals will have very "high risk" ratios... They mostly smoke too. No detectable CVD.

So I still remain a cholesterol sceptic. I keep thinking about exactly what is going on, but with Keys, Framingham and the MRFIT screenees as the basis of the lipid hypothesis, any link between cholesterol and heart disease will be an accident or a marker for other oxidative processes.


Stephan Guyenet said...


Just read your latest comment. Genius! I know you've put forth these ideas before in bits and pieces, but that's the most concise and complete version I've read. The theory explains why excess sugar is so toxic-- the fructose is a powerful glycating agent for ApoB100. And it explains why sugar and linoleic acid would be synergistic in CHD.

It also explains how plasma residence time of LDL could play into all this- oxidized ApoB100 extends it. It sounds to me like a theory that neatly integrates everything we know. I'm going to re-read some of your old posts on this. You should get yourself a lab and test some of these hypotheses. Come back to blogging soon!

Peter said...

Just editing

Peter said...

Hi Stephan,

The other side of the equation which keeps coming back to me is the role of omega 6 PUFA i insulin sensitivity. When I started looking at all PUFA I assumed that 3s would drop insulin resistance and 6s would increase it but this is not the case, certainly in the short term. So omega 6s acutely improve insulin sensitivity and omega 3s only do so in the much longer term, say over a year in humans. There is another EPIC Norfolk study which shows PUFA intake is inversely associated with HbA1c. In Norfolk this means omega 6s.

This looks to explain some of the switch between CVD deaths and cancer/violence deaths in MRFIT and various other pre statin lipid lowering trials.

Volek has a paper showing that LC ketogenic diets INCREASE arachidonic acid while decreasing every inflammatory mediator they looked at. We all know what mildly ketogenic diets do to HbA1c, so I suspect that non glycating diets might actually prolong LDL cholesterol persistence due to SFAs down regulating LDL receptors, while increasing the ratio of AA to EPA/DHA, yet still are therapeutic. High omega 6 PUFA diets increase LDL receptor numbers and shorten LDL persistence time, yes, and lower TC, yes, but they also lower HbA1c... Which is the beneficial effect on CVD?

This drives me back to Kwasniewski, who cares not about PUFA ratios while ever the carbs are low and saturated fats are high... Presuming there is a basic intake of DHA for essential needs like brains and eyes of course.

In Volek's paper the SFAs in the diet increase markedly yet there was no drop in the absolute amount of omega 6s due to the increase in total fat intake couple with adipose tissue sourced fat (weight loss).

This thought train makes me suspect that glycation, rather than absolute PUFA intake might matter more for CVD. Omega 6s become more significant for psychiatric and neoplastic diseases.

The other CVD and glycation aspect is the roll of hyperglycaemia in blocking the uptake of ascorbate in to vascular wall fibroblasts, giving failure to hydroxylate collagen and so worsening of pressure stress damage to arterial wall. Poor collagen requires oxidised LDL to apply a patch. LDL becomes sticky and oxidised under the conditions which inhibit normal arterial repair, just when it is needed. If you couple that with the damaging effects of hyperglycaemia on the glycocalyx of the arterial wall and the glycosaminoglycans of the ground substance of the arterial wall you have the situation where a change in diet gives an injury and the body has specifically evolved LDL to become sticky when sticking plasters are needed. Too much hyperglycaemia drives the repair process to produce a visible fatty natured disease, but the hyperglycaemia is the primary disease.

I remember Chris Masterjon's comment that traumatic injury to the arterial wall was not a model for CVD. Ascorbate deficiency damage in the Guinea pig is. It's a matter of what sort of trauma LDL is designed to respond to.

I feel Chris M is so right on so many things, but this framework outlined here is the one which keeps presenting itself to me. I'm still kicking at it, but it looks quite reasonable at the moment. These are all subjects for posts, about another 10 of them in all, which are due in to the AGE. RAGE and ALE series but I'll fill you in off blog about what's going on over here. I'm throwing this out unreferenced (as was the reply to Chris M's comment) in an hour's wait for the Building Regulations inspection on the insulation in the new bathroom. There are probably millions of typos.

And THINCS keeps throwing out little gems of papers that I should share.... Cracking intervention diet study from Glasgow. If you want to stay healthy stay well away from University dietitians, certainly in Glasgow! I'll have to go have a look.


Peter said...

BTW I came across a comment in Pure, White and Deadly where Yudkin claimed the flawed Finnish mental hospital study also fed less sucrose to their high PUFA, cholesterol lowering diet patients. Anyone have the full text to check this?????

OK, back to decorating.


Stephan Guyenet said...

Hi Peter,

I'm suspicious of the idea that linoleic acid improves insulin sensitivity in the long term. If you eat enough of it to give you fatty liver (which seems to be pretty easy if you combine it with sugar, given that 25% of Americans and British have it), it will certainly kill your whole-body insulin sensitivity.

The other thing is a highly controlled study I read a while back where they fed nearly 800 elderly men isocaloric diets that were either the standard 1960s American diet or one where the animal fat had been replaced by polyunsaturated vegetable fat. Over the course of 5 years, the PUFA group gained weight relative to the control group. That's not consistent with a long-term benefit to insulin sensitivity.

Back in the 1960s, the background LA intake was quite a bit lower than today so the intervention made a very large difference in their adipose LA stores and lipoproteins. The paper is free access, it's well worth your time:


Then there's the suppressive effect of LA on thyroid signaling in rodent livers. About the only way you can get a dog to develop atherosclerosis when feeding it saturated fat is by suppressing its thyroid.

I read that paper by Volek's group as well. It was counterintuitive to my simplistic view that n-6 derived eicosanoids are inflammatory, n-3 are anti-inflammatory. Well I've let that idea go. It seems like the bigger problem with excess LA isn't excess AA but the fact that it blocks n-3 elongation to DHA. DHA deficiency is exacerbated by increasing LA and has many negative consequences.

I've been trying to get information on CHD in the 19th century lately. They performed a ton of autopsies. It sounds like atherosclerosis didn't really get rolling until the late 19th century. But I still have some digging to do, so I'm not 100% confident on that yet. Angina was known back to the 18th C but seemed uncommon.

But here's the really interesting thing. Based on autopsies performed in the UK spanning the 19th and 20th C, atherosclerosis was decreasing as CHD deaths were increasing dramatically in the early part of the 20th C. People thought at the time, and I think it makes sense, that the real issue wasn't narrowing of the arteries but thrombotic tendency. That's something that's strongly affected by eicosanoids, and may represent the point at which LA exerts its most potent effects on CHD tendency. Clotting factor activity beats cholesterol as a CHD marker by a long shot. Just tossing out ideas.

Stephan Guyenet said...


If you haven't seen it, this paper is amazing. It's from St. Bartholemew's hospital in the UK. They show atherosclerosis and CHD death stats for autopsies performed 1867-1985. Atherosclerosis stayed the same, but CHD death rate skyrocketed even within age groups. Something like a 40-fold increase from the earliest time point to the 1950s. I think it's free access:


Charles R. said...

I just want to say that both you guys are doing great work here, and I know a lot of people are following this conversation with great interest.

The thing is that this kind of conversation, between you two, would either have been unlikely before the internet, or private. And like Gary Taubes points out, the ability to pull up all these studies on the internet just wasn't there before.

I think this significantly increases the velocity of this quasi-Socratic dialogue to discover the truth, but the vocabulary of the conversaton is the actual data. It's really fascinating to watch.

Mark said...

Finnish Mental Hospital Study.

Let's reevaluate the data and compare the low fructose Norm group with the high fructose Norm group.

And then, let's compare the low fructose SCL group with the high fructose SCL group.

Given food records, let's consider fructose as: sugar, sucrose, and fruit. Fruit is "berries and fruit" and this "includes dried fruits and preservatives.”

So you have one Norm diet in one hospital for 6 years. Then you have another Norm diet in another hospital once this one is over for another 6 years.

These are all in grams per day:

Low Fructose Norm Diet.
First 6 Years:
Hospital N: Norm Diet:
64g sugar.
78.5 sucrose.
97 fruits.

High Fructose Norm Diet:
Next 6 Years:
Hospital K: Norm Diet:
102 sugar.
102 sucrose.
140 fruit.

BIG difference in fructose.

Which had the better death rate?
Incidence per 1000.
Low Fructose Norm Group:
4.3 for death.
13.9 for "major ECG change or coronary death."
20.3 for "intermediate or major ECG change or coronary death."
Total is:38.5

High Fructose Norm Group:
28.3 (same order).
Total is: 47.8

The death rate was 9.3/1000 higher in the high fructose Norm group than in the Low Fructose Norm group.

Now for the SCL Groups.

Low Fructose SCL Group:
First 6 Years, Hospital K:
62g sugar.
68.4g sucrose.
74g fruits.

High Fructose SCL Group:
Next 6 Years, Hospital N:
87 sugar.
86.7 sucrose.
147 fruit.

Which had the better death rate?
Incidence per 1000.
Low Fructose SCL Group:
3.7 for death.
3.7 for "major ECG change or coronary death."
10.3 for "intermediate or major ECG change or coronary death."
Total is:17.7

High Fructose SCL Group:
16.8 (same order).
Total is: 23.8

The death rate was 6.1/1000 higher in the high fructose SCL group than in the Low Fructose SCL group.

So in both cases, the death rate went up with higher fructose intake.

BUT, the death rate, across SCL groups as explained by fructose, and across Norm groups does not explain the difference in death rate when comparing Norm Group with SCL Group.

Fructose on its own does not account for the increased death rate of the Norm vs. SCL group.

If you compare SCL Hospital K with Norm Hospital K, you find the slightly elevated fructose of the Norm group produces (assuming causality) extremely higher death rate.

Yet comparing one SCL group with the other, or one Norm group with the other, you find that a GREAT increase in fructose results in a comparatively small increase in death rate.

Here's another way of looking at it: The high fructose SCL group mortality vs. the low fructose Norm group mortality. High fructose SCL mortality: Total is: 23.8 per 1000.
Low fructose Norm mortality: Total is:38.5 per 1000.

So the high fructose SCL group has less deaths than the low fructose Norm group.

OK. What's the difference in fructose?

High fructose SCL:
87 sugar.
86.7 sucrose.
147 fruit.

Low fructose Norm:
64g sugar.
78.5 sucrose.
97 fruits.

So hands down, the SCL group has more fructose than the Norm group.

So even though the SCL group has more fructose, it has a lower mortality.


Chris Masterjohn said...

Hi Peter,

Thanks for responding. I'm sorry for taking so long to respond -- I forgot to subscribe to followup comments so didn't see your post until now.

I agree that it is possible to have a heart attack with TC below 150 mg/dL. My point was that the way you presented the Framingham data obscured the fact that no one in Framingham with TC that low had CHD.

I also agree that the role of oxidized LDL in promoting plaque formation is primarily an issue of PUFA toxicity and not total cholesterol levels in a causal sense, but correlations with total cholesterol levels reflect, in part, the causal role of this PUFA toxicity. Correlations never confirm or refute causation in the least, so it makes sense not to consider ourselves bound by the original causation hypotheses that unerlied the initial investigations into the correlations.

I agree that hemoglobin should not be glycated. But, glycated LDL is more likely to cause formation of plaque. A stronger correlation with blood sugar does not in any way suggest that the blood sugar comes first or that it is causally more important -- correlations do not show causation. Poor blood sugar control is both a cause and result of oxidation, so it is difficult to tease out sequence. It is also much easier to measure plasma indicators of. Oxidation of LDL, for example, leads to accumulation of oxLDL in macrophages or uptake into the liver -- where is the plasma indicator?

I agree that low-carb reduces inflammation. Studies have been done in my department (UConn) showing low-carb to have vast superiority to low-fat in reducing inflammation and treating metabolic syndrome.

I commend your ability to look at the big picture in FH. But vitamin A is transported by RBP/transthyretin, vitamin D is transported by DBP, vitamin E is transported to endothelial cells by HDL and dependent on SRB1 for selective uptake, not dependent on lipoprotein uptake. Vitamin K and coQ10 might be compromised in FH, but lipoprotein lipase is probably a much more significant contributor to peripheral cell uptake of these vitamins than LDL-R. Most LDL-R is expressed in the liver, so quantitatively, it's primarily affecting the concentration and residence time of LDL in the blood.

More importantly, the initial stage in the development of atherosclerotic plaque, at the point where we can agree that a plaque is beginning to exist, is the formation of foam cells. So the question we should ask is, how does FH increase foam cell formation? And of course it primarily comes down to oxidized LDL, which inhibits NO and increases cytokines that draw monocytes, but much more importantly causes monocytes and macrophages to differentiate into foam cells.

The Brown and Goldstein quote is not contradictory to a central and necessary role of oxidized lipoprotein PUFAs in atherosclerosis. It just indicates the diesease is multi-factorial, which was clearly established as early as the cholesterol-fed rabbit model.

LDL-R activity can guarantee heart disease in infancy or nearly abolish it (88% risk reduction) -- you don't see it is a main factor governing atherosclerosis risk?

I agree that LDL-R is no the *only* factor affecting risk -- but the totality of activity includes much more than cell surface concentration. If SFA reduces liver cell surface LDL-R but also reduces ApoB oxidation, then it could, net, increase interaction of LDL with LDL-R, thus effectively causing an increase in LDL-R activity. Moreover, fatty acids affect thyroid function, which is another main governor of LDL-R activity.

The Kitavans are a separate population and one cannot use risk marker levels developed in different populations from epidemiological studies as surrogate markers to ascertain their LDL-R activity.

Keys, Framingham, and MRFIT are all epidemiological studies (at least Keys' most commonly cited work). These cannot be used to infer causation, whether for or against. Besides, historically, the basis of the lipid hypothesis was the cholesterol-fed rabbit model. There is an enormous amount of experimental data we need to look at too.


Peter said...

Hi Chris,

The main reason I look at glycation of Apo B100 being a significant first step in the oxidation of the lipids in LDL is this paper here:


To me the occurrence of a glycation hotspot, which essentially turns off apoB100's attachment to the LDL receptor and at the same time oxidises the linoleic acid


in the particle to a set of macrophage chemotactic substances is not random.

In the larger picture, the presence of glycating conditions sets up persistent glycated and oxidised LDL in the bloodstream. The organisation of LDL in this way cannot be a recent evolutionary phenomenon, so I assume it has a survival advantage.

Glucose damages endothelial cells


so I again feel that hyperglycaemia does the damage and the LDL particle is set up to be affected by this hyperglycaemia and becomes modified appropriately. It is designed to glycate.

As IHD appears to be a recent phenomenon and chronic hyperglycaemia is also a novel phenomenon, my hypothesis is that LDL is designed to glycate/oxidise to deal with the effects of transient hyperglycaemia. Chronic hyperglycaemia is the novel pathology which leads to the development high levels of oxidised LDL and subsequent atheroma. To me the glycating environment is what comes first.

Of course hyperglycaemia activates NFkappaB in the liver with all of the downstream pro inflammatory effects this entails




I don't think anyone would argue with the premise that IHD is an inflammatory condition. A simple glucose infusion is a powerful tool to set the stage for IHD. I see small dense LDL as glycoxidised particles produced in response to the hyperglycaemia. Anecdote form sources such as Dr Davis' site suggest that THE tool to reduce or eliminate small dense LDL (probably the glycoxidised LDL particles going nowhere other than in to a macrophage) is the elimination of refined carbohydrate, especially sugar and wheat products. HbA1c would be an excellent long term surrogate for measuring sdLDL and a good predictor of IHD. Of course Dr D is a full lipid believer so does not seem to track glycation, other than indirectly through NMR to pick up sdLDL.

On the other hand the lipid hypothesis initially identified TC as THE marker of heart disease risk, and saturated fat as THE elevator of TC, if I am reading Keys correctly (I believe he worked on the basis that dietary cholesterol was unimportant, so quite where egg phobia came from goodness only knows). For this to hold water you then have to accept TC was a surrogate to LDL and now we have to accept that LDL is a surrogate for sdLDL and that sdLDL is pathological in its own right. It feels like a house of cards to me compared to the glucose hypothesis, but that probably reflects my own basic biases.

Which are quite strong.

I appreciate the info on vitamin transporters I'll take that on board. I come back to the problem in FH being a lack of LDL receptors and that actively growing vascular endothelial cells up regulate the number of their LDL receptors.


I have to ask myself why these cells want to endocytose LDL and if there are negative consequences of their not getting it or getting it in short supply. I then come back to Brown and Goldsteins ideas for managing FH. They do nothing to help the endothelial cell when they up-regulating hepatic LDL receptors and drop serum LDL levels. They may effectively treat a lab number but once you start screening the general population you will end up treating many people who do not need treatment, the healthy FH patients. As you are no doubt aware heterozygous FH patients have remarkably normal life expectancies unless they are pre selected from families where FH co exists with IHD. In the Dutch and Utah pedigree studies of IHD-free heterozygous FH individuals there is only a transient mild increase in all cause mortality risk and that only occurred in the mid 20th century.


On the infants with IHD front I have seen the abstracts of several case reports but not the reports in detail. If the absolute level of LDL is the explanation of the need for heart/liver transplants in these children I would expect that the problem would be ubiquitous in FH homozygotes at similar levels of LDL, but case reports of children with comparable LDL levels without premature CVD don't seem to be available. This may reflect the possibility that all children of comparable LDL levels have comparable problems. Perhaps so but, as always, those without problems, the healthy FHers, may not get in to the literature. So on this front I just don't know. Is there a level of LDL which automatically causes IHD in human infants?

OK, I think that's probably enough for now...


Peter said...

Hi Stephan,

Thanks for the links, some rethinking and interesting reading to do there. The THINCS members all have their own ideas about IHD and there is some body of evidence to suggest even the clotting is secondary. But of course if there is a primary injury adding anything to the cascade would plausibly worsen the outcome. There is just too much to read. Having said that acute hyperglycaemia is a good way of upregulating your clotting system... And also your sympathetic nervous system level of activity. So I stay gluocentric in my thinking.

And of course sudden cardiac death is usually related to ventricular fibrillation rather than the actual dead cardiac muscle mass, so omega threes come in again.

Anyway I still have a lot of thinking to do! I hope you'll be distilling the atherosclerosis time data for us... A quick skim of the paper looks fascinating.

BTW I have papers somewhere on my hard drive about naturally occurring IHD in both rats and rabbits kept to old age. They certainly do get IHD but not arteriosclerosis. Their lesions resemble very early ground substance damage and fibrosis seen in humans (including neonates) but not atheroma development. I have those to think about too...


Peter said...

Hi Mark,

Thanks for doing the math. There goes another idea down the drain!


Chris Masterjohn said...


Thanks for responding!

I'm sorry that my reply will be briefer and less referenced than is justified by this discussion but I am swamped this week.

I agree that hyperglycemia, if present, will contribute to atherosclerosis by increasing glycation of LDL.

However, from what I have read, it is believed that LDL membrane PUFAs are the first target of oxidation and the ApoB follows. The evidence for this, as I understand it, is that if you segregate LDL from plasma according to its degree of oxidation, the least oxidized LDL has only its PUFA oxidized, whereas LDL in more advanced states of oxidation has its ApoB and in some cases cholesterol oxidized.

The term "glycation" is deceptive and should be changed. Free PUFAs are vastly more effective than glucose in inducing glycation, even in the absence of any glucose at all. Hyperglycemia is not necessary for a pro-glycation environment.

I do not consider it very plausible that LDL is designed to be a target of glycation. We are not designed to have hyperglycemia, PUFA overload, nutrient deficiency, heavy metal toxicity, or other conditions that contribute to this process.

When LDL becomes oxidized/glycated, it becomes pathological and atherogenic. We are not designed to have atherosclerosis.

I do think the macrophage is designed to mop up toxic factors such as oxidation products, and I do think its role in atherosclerosis is to protect against the greater evil of having those oxidation products damage endothelial cells.

We seem to be in agreement on this point. I just don't think that the hot spot on ApoB is part of the mop-up system any more than I think endothelial cell DNA is there to mop up oxidation products by falling apart. I think the purpose of ApoB is to transport lipids efficiently. When they are transported efficiently, the oxidation doesn't happen.

I don't see how Keys can possibly be credited with originating the lipid hypothesis. This is owed to Anitchkov and the cholesterol-fed rabbit. Anitchkov very clearly identified the role of inflammation in this process decades before Keys made any contributions, and himself said that dietary cholesterol was not the initiating factor in the human.

I do not see why you see glucose as an alternative to lipid. Glucose promotes lipid oxidation, free PUFA promotes lipid oxidation much more effectively, transition metals catalyze lipid oxidation, acidic environments free transition metals to allow this catalysis, thyroid hormone clears lipids from blood and thus prevents their oxidation, exercise seals the arterial wall from lipid oxidation products -- the center of all this is lipid oxidation, and glucose is, like all these others, a contributing factor.

Your analysis of FH is very unfair. Obviously, if FH contirubtes to increased mortality through increasing the chance of premature atherosclerosis, then if you isolate the subpopulation that does not have premature atherosclerosis, you will have normal lifespan! The point is that heterozygous FH have many times the risk of premature atherosclerosis as those who do not have FH!

And no, you would absolutely not expect homozygous infants to have uniform risk of CHD if they have univform lipid levels. Because CHD is multi-factorial!

However, as Anitchkov's research and the many other experimental models of atherosclerosis in animals showed, reviewed on my site (www.cholesterol-and-health.com), hyperlipidemia, which is really a state of hyperoxidizedlipoproteinemia induced by suppression of LDL receptor activity and decreased clearance of blood lipid, is essential to the proces of atherosclerosis.


Stephan Guyenet said...


I'll be posting on the autopsy paper eventually. I'm not totally convinced that thrombosis is the answer either, but it's an interesting alternative to consider. I also think it's interesting that Thais eating white rice got very little atherosclerosis in the 1960s when Brits have had atherosclerosis eating white flour since the 1860s.


The point Peter was making about FH is that in the Dutch study (the one I sent you a few weeks ago), the hets did not have an increased risk of CHD or total mortality. You don't get that in the abstract, you have to read the full text. The difference between their study and others is that they identified FH hets by completely random routine cholesterol screening, rather than through family relation to another person who already had CHD.

When they compared FH hets who were selected by random screening to other Dutch selected by random screening, the risk was the same. When they compared FH hets selected on the basis of being related to CHD patients, their risk was elevated but the same as non-FH hets selected in the same way. They suggest (very obliquely, in the discussion section) that other studies may have gotten it wrong due to failing to use an appropriate control group.

Here's how the data could get confounded. Patients with CHD risk factors like overweight, diabetes and smoking get screened for high cholesterol more than the general population. During the course of this screening, FH homos and hets will be discovered. Therefore, the pool of identified FH hets is weighted toward relatives of people who have CHD risk factors. If you then compare those people to the general public, they will not surprisingly be at an elevated risk of CHD. What the Dutch study suggests is that if you use an unbiased selection procedure, the difference disappears.

Peter said...

Hi Chris,

I know what you mean about things to do!

The sequence of changes in glyco-lipoxidation (perhaps call it "degroxidation" to avoid blaming a specific substance?) of LDL is new information to me and, as it should do, it gives me more information to fit in to my ideas of how things work, but it does not alter my personal view that the degroxidation of apoB100 diverts LDL particles from uptake in to non immune cells to uptake in to macrophages. It does mean that I need to consider what this suggests in context, but certainly does nothing to alter my basic concept of a switch on apoB100. I am also convinced that apoB100 is part of the innate immune system and, as such, has functions in addition to lipid transport.

Back in the days when Bruce K considered me worthy of his time, I was informed that PUFA were excellent "degroxidation" agents. To me the obvious follow on from this is that simply replacing saturated fat in the diet with PUFA in the diet would result in a rise in HbA1c in the absence of any change in blood glucose. My attempts to locate such data have been unsuccessful and certainly Bruce K never came up with any. In fact, the EPIC Norfolk dietary fat type vs HbA1c paper, while very third rate and observational in nature, suggests the direct opposite (probably through confounding factors, but their data certainly do not suggest dietary PUFA as glycators of HbA1c). Where as eating LC, without changing total PUFA intake, will automatically drop HbA1c, in proportion to the degree of reduction in glycaemia. Am I missing core studies here?

I think we are destined to disagree about whether LDL degroxidation is pure pathology or adaption driven to pathology by the factors you mention. This is fine by me. If everyone agreed, we would have all of the answers. We're not in that Nirvana!

My analysis of FH is certainly not intended to be fair. There are sufficient case reports of FH in CVD prone families without any consideration, beyond cholesterol levels, of what might be going on that a little extreme the other way does not seem like a bad thing to me. There certainly are FH families with minimal cvd and also, with advancing age, heterozygous FH becomes progressively protective against CVD.

These things do need thinking about. I find them interesting.


Chris Masterjohn said...


I see something very different when I look at the study.

They had 113 families referred as FH and looked at all first-degree relatives, half of whom have FH, so the excess mortality represents half of the excess mortality due to FH. The FH population studied had a 34% excess of mortality indicating a 68% excess of mortality due to FH.

Then, post-hoc, they divided this population into two subpopulations, one diagnosed on the basis of CAD and the other without CAD present. Thus, they introduced a bias into both subpopulations: they separated them according to whether a relative had premature CAD, indicating the probable existence of additional risk factors in that person. It seems obvious to me that the subpopulation without CAD is not in any way a random sample of the FH population; rather, it is a specific subpopulation that does not have CAD.

The authors do not seem to interpret their data the way you do. "Presumably these families (the older studies) -- and our families with a premature onset of CAD -- were characterised by clustering of risk factors. In the present study, a lot of families were not ascertained by clinical outcome but by routine measurement of cholesterol and they had a life expectancy similar to the Dutch population, suggesting the presence of protecting factors or the absence of additional risk factors. . . . Many carriers of mutations in the low-density lipoprotein receptor gene reached a normal life span suggesting that untreated hypercholesterolaemia itself does not always cause excess mortality."

If the risk factors are "additional," then they consider FH to be a risk factor. If FH does "not always cause excess mortality," then it sometimes causes excess mortality. What they are saying is that, while FH in and of itself contributes to heart disease, there are unknown additional factors that aggravate or protect against the risk, causing two subpopulations of FH to emerge: those at risk for excess mortality due to lack of protective factors or presence of additional risk factors, and those not at risk for excess mortality due to the inverse situation.

To say that not everyone with FH gets premature heart disease, therefore FH makes no causal contribution to heart disease, is not sound logic.


Chris Masterjohn said...


Thanks for your response. I'll have to look up the structure of Hba1c to make sure, but my understanding is that it is glycosylated. PUFAs contribute to glycation, a much broader process, more effectively than glucose in the complete absence of glucose. In other words the process of "glycation" does not necessarily involve any sugars. So I think the premise that by decreasing glycation of LDL, not necessarily involving sugars, you will ipso facto increase glycosylation of hemoglobin, involving sugars, is incorrect.

Moreover, oxidative stress contributes to blood sugar dysregulation, so if it is true that PUFAs contribute to oxidative stress, reducing them would increase blood sugar control. Indeed, in the alloxan model of diabetes, PUFAs increase the diabetogenic effect of alloxan and saturated fat protects.

I guess we share different views of the role of bias. Everyone is bound to have their own biases, but I believe we should constantly strive to keep our own biases in check, to search for them and route them out wherever we find them.

I think in the case of FH you are conflating correlation with causation. FH is inversely correlated with CHD in older age groups -- that does not mean it becomes "protective." Of course, it may be, but that is a hypothesis one has to argue as biologically plausible and then seek to develop evidence in confirmation of, not something one can simply infer from correlation.


Peter said...

Hi Chris,

From my point of view, if you glycosylate haemoglobin you will glycosylate apoB100. Both will occur using a Schiff base and glucose and if it is happening to Hb it is also happening to apoB100. You are welcome to free radical generated damage to (glycation) lipids, by lipids, (in the absence of glucose) before this occurs, so be it as you have the data.

However I don't see any evidence of a mechanism by which glycation will prolong LDL residence in serum, but clearly glycosylation will do this.

The best way of up regulating LDL receptors in a person without other medical issues is to increase the linoleic acid in their diet. The best way to increase PUFA in LDL is to do the same. Is this good? Saturated fats will down regulate LDL receptors, prolong residence time of LDL in plasma yet minimise progression of arteriosclerosis, provided they replace carbohydrate.

Low carbohydrate diets will decrease inflammation, decrease cv disease incidence and lower HbA1c without any requirement for decreased PUFA in the diet.

I don't like sunflower oil, I find it weird to be defending it. But I can't see any logic to your position.

If you had come back to me with PUFA plus sucrose (as did Stephan) causing hepatic insulin resistance to affect HbA1c I'd have been interested. But this would be glycosylation, not glycation I guess.

The alloxan diabetes issue is an aside. I assume you are talking about the Rodriguez paper (http://www.ncbi.nlm.nih.gov/pubmed/14933592) on the induction of diabetes, otherwise any diabetes model would do. How about the contradictory findings:



I would say that I dont have the full text of these papers, but if Rodriguez is your citation I would be interested in whether palmitic acid was as protective as caprylic acid. This has some influence on what is happening.

Back to cardiovascular disease in humans, my standpoint:

Normoglycaemia WITHOUT instruction to reduce PUFA markedly reduces cardiovascular incidents

Eating less carbohydrate and more saturated fat WITHOUT any less PUFA is associated with less CVD progression

Reducing carbohydrate WITHOUT reducing PUFA decreases inflammation

Why am I glucose fixated?

Because normoglycaemia looks to me to be what matters in avoiding CVD progression. You're not convincing me otherwise so far, any more than Bruce K did.


Chris Masterjohn said...

Hi Peter,

Oxidation of LDL, or many other forms of modification that have been studied, some of which do and some of which do not occur in vivo, prevent LDL from being taken up by LDL-R, thus, in vivo, prolonging residence time in plasma.

There are a number of issues you bring up that I need to study further before commenting, and this week is not a week I can do that, but we can chat more in the future.

Thanks for engaging my comments,

Peter said...

Hi Chris,

I disappear off line, as of about now, for the next 10 days or so...


Stephan Guyenet said...


I've taken another look at the Dutch FH paper and I've decided that neither of us is interpreting it correctly.

I believe your point was that the FH families selected on the basis of not having CHD are not a fair comparison with the general Dutch population. Your point is well taken.

However, the paper states that "In 62 out of 113 [FH] families the referral of patients was based on the presence of premature cardiovascular symptoms in at least one family member". This is not an unbiased selection of FH patients and thus it can't be compared to the general Dutch population either. So the increased likelihood of CHD and overall death they found is difficult to interpret.

What you would need to get to the bottom of this question is a sample of FH hets who had been selected by random screening rather than referral through a moribund relative, and compare them to a random sample of the population.

Chris Masterjohn said...


I agree with you, and I think that is basically what I said, that you're looking at two separate subpopulations of FH. The question is whether the subpopulation with premature CHD is overrepresented compared to the subpopulation with premature CHD in the general population. The general consensus is that it is overrepresented five-fold, but I'd have to give more thought to your proposal about how to properly measure it as well as read the primary research out there, which I have not done on this particular point, before coming to a solid conclusion.


Stephan Guyenet said...


OK, it sounds like we're on the same page. I wonder if anyone has been able to look at FH hets in a truly unbiased manner?

Ace said...

Interesting study:


They found that RBCs live longer in people with good blood sugar control; longer life means more time to be glycated.

It seems like this means that HbA1c isn't as good of an indication of average BG levels as some people think. Low BG plus long RBC life could still result in moderate HbA1c. Conversely, high BG with short RBC life could result in low HbA1c.

Peter said...

Hi Ace,

Absolutely. The classic example is Stephan at Wholehealthsource who has an HbA1c of the mid 5%s but never cracks 5mmol/l BG post prandially and runs in the low 4s mmol/l pre prandially. I would also suggest low PUFA in the erythrocyte membranes increases erythrocyte survival time, depending on what the immune system looks at to determine death time for RBCs.

A low HbA1c from a GI bleed should be bad news but iron depletion protects against CHD in iron storage diseases and possibly within the normal population... There are a fair few testable hypostheses generatable from EPIC!


Ace said...

What does this imply about the reduced risk of CHD episodes with very low HbA1c? If HbA1c only declines to a certain point with low BG, then what is providing the additional benefit (if there is one)?

Or maybe we're not seeing the whole story?

I remember an interesting study where they looked at RBC zinc levels as a function of age. The researchers expected levels to decline with age; what they found was the reverse. In analyzing the data, what they found was that people with low zinc died younger (the zinc level and age axes weren't independent).

Could something similar be happening here? Is it possible that the reason why there's less CHD in people with ultra-low HbA1c is because they are dying early from other causes?

Peter said...

That's always the way with observational studies. As far as I am aware there have been very few interventional studies where an attempt was made to lower HbA1c in type two diabetics. The couple of mega money waste studies in the USA simply added polypharmacy to a sugar based diet so tells us nothing about HbA1c and CVD risk, or all cause mortality for that matter.

The nearest I've seen is the small study by Nielsen in Sweden. Too small for all cause mortality but the soft CV event rate is lowered by LC associated with lower HbA1c (combined with reduced meds). Of course it might be the reduced meds, not the reduced HbA1c, that reduced soft end points...


. said...

In the Framingham Heart Study there was no association between cholesterol and CHD. Here are the histograms of original data in Excel format http://bit.ly/nRwn22