Sunday, September 30, 2012

Lipids, Lipoproteins or Function: HDL

HDL Cholesterol and Particle Number

HDL-C is a risk factor for CVD, but in a previous post I showed examples (niacin and CETP inhibitors) where increasing HDL-C didn’t reduce CVD.  To really drive home the point here’s a study (the same one as the last post) that measured the HDL-C levels of people hospitalised for CVD.  They found that almost half had an HDL-C above 40 mg/dl


As I said before, this is an observational study, so we can’t establish cause and effect and it’s subject to confounding variables.  Also, to use this data to draw a solid conclusion it would need to be compared to population levels, otherwise one could look at this data and conclude that ‘HDL-C levels of 0 are associated with a reduced risk of CVD’.  What I want to show though, is that having a high HDL-C certainly doesn’t make you immune to CVD

Since LDL-P is more strongly associated with CVD than LDL-C, researchers have investigated the relationship between HDL-P and CVD and found that HDL-P is a better risk factor than HDL-C [2]

HDL Function

But the reasons why HDL is considered protective is not some linear relationship with the amount of cholesterol or the particle number, but rather the functions of HDL.  HDL function is more strongly negatively associated with CVD and HDL function doesn’t depend on HDL-C or ApoA-1 and is not associated with either, which explains the large variation in HDL-C levels in those hospitalised for CVD [3].

HDL has several functions that are considered to be protective including reverse cholesterol transport, anti-oxidative and anti-inflammatory effects, and anti-apoptotic, vasodilatory, antithrombotic, and anti-infectious activities.  HDL loses these functions and becomes ‘dysfunctional’ under conditions of oxidative stress, inflammation and infection.  High triglycerides also impairs HDL function.  Serum amyloid A (SAA) and myeloperoxidase (MPO) (two inflammatory molecules) are strongly associated with CVD and inhibit HDL function.  People with dyslipidemia or the metabolic syndrome have impaired HDL function and small dense HDL particles.  Notably, in people with the metabolic syndrome their HDLs are less effective at inhibiting LDL oxidation.  By the way, dysfunctional HDL particles are usually small and dense [4]

While a pro-inflammatory environment impairs HDL function, an anti-inflammatory environment promotes HDL function.  Regulatory T cells/IL-10 increase cholesterol efflux to HDL and reduce cholesterol uptake by macrophages, thereby safely dealing with oxLDL and oxidised cholesterol but also reducing atherosclerotic plaque [5] [6].  Evidence supporting the role of regulatory T cells/IL-10 in CVD:

  • People with CAD have about half as many Treg cells and their Treg cells have impaired functioning [7] [8]
  • Overexpressing Treg cells [9] and IL-10 [10] [11] reduces atherosclerosis in an animal models of CVD
  • IL-10 knockout mice are more susceptible to atherosclerosis [10]
  • IL-10 is associated with better CV outcomes [12]

However, one confounding variable is that oxLDL promotes apoptosis of Treg cells [13] and IL-10 production [14].  A polymorphism that decreases IL-10 by 30% isn’t associated with atherosclerosis [10] and a study finds Treg cells are not associated with atherosclerosis [15]

* The mechanism by which vitamin D is protective against CVD may be by improving Treg cell function

Sunday, September 23, 2012

Lipids, Lipoproteins or Function: LDL

LDL Cholesterol

LDL-C is a risk factor for CVD, but in the previous two posts I showed examples where lowering LDL-C didn’t reduce CVD.  To really drive home the point here’s a study that measured the LDL-C levels of people hospitalised for CVD.  They found that almost half had an LDL below 100 mg/dl and 17.6% had an LDL below 70 mg/dl


This is an observational study, so we can’t establish cause and effect and it’s subject to confounding variables, such as statin use for secondary prevention or people with high risk.  Also, to use this data to draw a solid conclusion it would need to be compared to population levels, otherwise one could look at this data and conclude that ‘LDL-C levels of 220 are associated with a reduced risk of CVD’.  What I want to show though, is that having a low LDL-C certainly doesn’t make you immune to CVD, far from it in fact

* Their conclusion is very different to mine “These findings may provide further support for recent guideline revisions with even lower LDL goals” [1]

The LDL Particle

But of course, why would we expect LDL-C to be definitive.  In atherosclerosis I mentioned the response-to-retention hypothesis and the oxidative modification hypothesis.  Both hypotheses relate to the LDL particle: whether it’s the LDL particle penetrating the endothelium and then getting stuck (response-to-retention); or the LDL particle’s proteins becoming oxidised (oxidative modification).

The rest of this post will briefly cover the response-to-retention hypothesis, which is discussed more in detail in the first three links at the end of the post.  But it’s equally important to remember that LDL oxidation is necessary for atherosclerosis.  I’ll mention two causes of LDL oxidation in later posts

LDL Particle Size

One explanation for variation among LDL-C in CVD is the size of the LDL particle.  There are two types/sizes of LDL: pattern A or large and buoyant (lbLDL); and pattern B or small and dense (sdLDL).  sdLDL is associated with increased triglycerides and decreased insulin sensitivity and HDL-C.  sdLDL are more vulnerable to oxidation and increase the risk of CAD by 200% [2]

The proposed mechanism is that the sdLDLs are small enough to get stuck between the endothelial cells, where they are exposed to more oxidants directly from smooth muscle cells and endothelial cells and are exposed to fewer antioxidants from the bloodstream and spend more time in the bloodstream, which increases LDL antioxidant depletion

Particle size is often used as an argument for low carb diets as diets lower in fat and higher in carbohydrates increase the number of people who predominately have sdLDL particles.


LDL Particle Number

However, sdLDL is associated with ApoB, which is a markers of LDL particle number (LDL-P).  More importantly the risk of sdLDL in CVD is dependent on ApoB, which means its value as a risk factor is abolished after adjusting for ApoB* [2] [3].

The proposed mechanism for sdLDL also has some problems.  Modified LDL is denser as the phospholipids leave the LDL particle and go to an HDL particle, and sdLDLs have a different oxidation state to regular LDLs.  So it seems like LDL oxidation precedes sdLDL [4].

LDL-P is a very a strong risk factor for CVD (stronger than LDL-C) and remains one after adjusting for sdLDL.  High LDL-P is associated with the metabolic syndrome, low HDL-C and high triglycerides.

The suggested mechanism for LDL-P is similar to sdLDL, which suggests the more LDL-P, the more likely LDL particles are to penetrate into endothelial cells, which then initiates an inflammatory response and produces atherosclerosis.  Two other factors in this model include the permeability of endothelial cells and a maladaptive immune/inflammatory response [5]

* Please don’t use this study to promote low carb diets while simultaneously ignoring LDL-P

LDL, the Thyroid and CVD

The LDL receptor is found mostly in the liver and allows LDL-C and LDL-P to be cleared from the bloodstream.  In familial hypercholesterolemia (FH) the LDL receptor is defective or absent, so LDL particles remain in the bloodstream for longer (~2.5 vs 4.5 days) and therefore LDL-C and LDL-P are elevated.  The opposite occurs with some mutations of the PCSK9 gene, where the LDL receptor is more active, LDL-C and LDL-P remain in the bloodstream for a shorter period of time and are lower.  People with FH have more CVD and people with the PCSK9 mutation have less CVD [6]

Having downregulated LDL receptors isn’t ideal: the response to retention hypothesis suggests that a higher LDL-P in the bloodstream for longer increases the number of LDL particles that will penetrate the endothelium.  The oxidative modification hypothesis suggest that the longer LDL particles remain in the bloodstream the more their antioxidants will become depleted and which can lead to the LDL particle becoming oxidised

Besides gene mutations and statins, thyroid hormone also influences the activity/concentration of the LDL receptor.  Hypothyroidism is often, but not always, associated with higher total-C, LDL-C [7] [8] and atherosclerosis/CVD [9] [10] [11].  Giving thyroid hormone to normalise low metabolic rate reduced heart attacks by 76% in old men and 85% in young men [12].  For a in depth look at the thyroid and CVD see the last link

Further Reading:
(1) The straight dope on cholesterol – Part IV
(2) The straight dope on cholesterol – Part V
(3) The straight dope on cholesterol – Part VI
(4) Thyroid Hormones and Heart Disease

Sunday, September 16, 2012

The Lipid Hypothesis: Blood Lipids and Ratios

Reverse Cholesterol Transport and The Total:HDL-C Ratio

Total cholesterol is an old risk factor for CVD and is quite an odd risk factor as it’s made up of cholesterol in LDL (LDL-C), HDL (HDL-C), IDL (IDL-C) and VLDL particles (VLDL-C), with LDL-C, IDL-C and VLDL-C being considered bad and HDL-C being considered good.   Surely if you consider LDL-C bad and HDL-C good then you should look at them separately or a ratio of them.

The reason HDL is thought to be good is because it transports cholesterol back to the liver (reverse cholesterol transport) and has antioxidant and anti-inflammatory effects [1]. 

The process of reverse cholesterol transport goes like this: 
  1. HDL transports cholesterol out of cholesterol rich macrophages (foam cells).
  2. Cholesterol ester transfer protein (CETP) exchanges the cholesterol in HDL particles with triglycerides in VLDL particles.  This lowers HDL-C and changes the VLDL particles into (small, dense) LDL particles
  3. The triglycerides in the HDL particle are degraded to fatty acids and glycerol via hepatic lipase, while the cholesterol in the LDL particle is transported to the liver and taken up by the LDL receptor often to be used for bile production 

A high Total:HDL-C ratio suggests the process of using and clearing cholesterol is inefficient, and is a stronger risk factor than total cholesterol, LDL-C, HDL-C or triglycerides.

“So what would the total-to-HDL cholesterol mean? The longer LDL stays in the blood, the more two things happen: it is exposed to oxidants, and as its limited supply of antioxidants run out, the polyunsaturated fatty acids in its membrane oxidize, leading to the further oxidation of its proteins and cholesterol; it is exposed to cholesterol ester transfer protein (CETP), which transfers cholesterol from HDL to LDL, thus boosting the total-to-HDL cholesterol ratio.” - Chris Masterjohn [1]

OR (10% increase)


A meta-analysis of cohort studies found that "the ratio total/HDL cholesterol was the strongest predictor of IHD mortality (40% more informative than non-HDL cholesterol and more than twice as informative as total cholesterol)" [4]

The Total:HDL-C ratio seems to be a pretty good risk factor (much better than total cholesterol) and the lipid hypothesis should probably be modified to suggest that: 'reducing the total cholesterol: HDL cholesterol ratio in an individual or in a population group will lead to a reduction in the risk of suffering a new event of coronary heart disease'

* In the second reference the likely reason why the Total:HDL-C ratio had a higher risk than the LDL-C:HDL-C ratio is because total cholesterol also includes IDL-C and VLDL-C (which may be worse than LDL-C, not sure).  non-HDL-C would be better than LDL-C to use for this reason.  Also, because total cholesterol and HDL-C is measured in standard blood tests you don't have to make approximate calculations that may or may not be accurate.

Testing the Modified Lipid Hypothesis

Once again, this modified lipid hypothesis is still a hypothesis.  While it may be true most of the time it still needs to be tested to see whether things that lower/raise the ratio actually decrease/increase the risk of CVD.  Previously I mentioned that phytosterols, replacing SFA with MUFA/PUFA or carbs, grain fibre and low fat diets reduce total cholesterol but don't reduce CVD.  Because these things reduce LDL-C proportionally more than HDL-C they also reduce the Total:HDL-C ratio, thus once again showing the importance of testing hypotheses and not assuming that just because something reduces the Total:HDL-C ratio it will therefore reduce CVD.  In addion, niacin and CETP inhibitors reduce the Total:HDL-C ratio, but don't reduce CVD

Niacin is vitamin B3, but when used in pharmacological doses (~100x the RDI) it’s considered a drug.  Niacin lowers total cholesterol, LDL-C and triglycerides and raises HDL-C, thereby lowering the Total:HDL-C ratio.  But niacin doesn't reduce CVD and may increase the risk of stroke [5]

CETP has gained some attention and is considered to be atherogenic because it generally increases LDL-C and decreases HDL-C, therefore increasing the Total:HDL-C ratio.  However, the relationship between genetic variations that reduce CETP expression and rates of CHD is inconsistent, some variants increase the risk, while others decrease the risk [6].  For example, one study found people with an HDL-C of 41-60mg/dl and a heterozygous mutation for CETP deficiency had 43% more cardiac events.  The authors of the study described CETP deficiency as “an independent risk factor for CHD” [7].  Although this hasn’t stopped pharmaceutical companies from developing CETP inhibitors.  In RCTs CETP inhibitors have resulted in more deaths than placebo.  This is probably because HDL cholesterol doesn’t protect against atherosclerosis, HDL particles do, and a way in which HDL particles protect against atherosclerosis is by carrying vitamin E and CETP transfers vitamin E to HDL particles [8] 

Further Reading:
(1) The Total-to-HDL Cholesterol Ratio -- What Does It Mean?

Sunday, September 9, 2012

The Lipid Hypothesis: Total Cholesterol

The Lipid Hypothesis 

The lipid hypothesis is based off the Framingham study, which found that the higher one’s cholesterol, the greater their risk of cardiovascular disease.  The lipid hypothesis suggests that: 

“Reducing the level of plasma cholesterol in an individual or in a population group will lead to a reduction in the risk of suffering a new event of coronary heart disease” [1] 

Given the evidence from Framingham it’s not an unreasonable hypothesis.  Though it’s important to remember that the lipid hypothesis is just a hypothesis, an educated guess of what may happen when you lower someone’s cholesterol.  If the relationship between cholesterol and CHD remains consistent, things that lower cholesterol will still need to be tested to see whether they actually reduce events of coronary heart disease (as correlation doesn’t prove causation). 

So two things need to happen: (1) there must be a positive association between cholesterol and CHD events; and (2) things that lower cholesterol need to be tested to see if they actually reduce CHD events
Total Cholesterol and Mortality

Total cholesterol was the only measurement of blood lipids to begin with and continues to be the most common risk factor that is looked at for CVD.  But we shouldn’t just look at associations between cholesterol and CVD.  If the aim of reducing CVD events is to live longer than we should also look at the relationship between cholesterol and all-cause mortality as well.

Meta-analysis of cohort studies (aged 35-69) [2]
(other includes non-cardiovascular and non-cancer causes such as respiratory, digestive, trauma, etc)
All-cause mortality in a review of observational studies and RCTs in people aged 80+[3]
Using WHO data to compare countries based on average cholesterol and mortality [4]
Of the studies I would put more weight in the first one as the second one looks at only a small section of the population and the third is subject to more confounders (there are many differences between countries).

In summary, the relationship between total cholesterol and CVD is generally positive, whereas the relationship between total cholesterol and all-cause mortality seems to follow a U-shaped curve where cholesterol levels between 160-239 mg/dl (4.2-6.3 mmol/l) are associated with lower mortality.  This is because low cholesterol is associated with a higher cancer and other causes of mortality in the first study and infectious disease in the third.  Based on this data, high cholesterol should be defined as > 240 mg/dl (> 6.3 mmol/l), but even still, the increased risk at this 'high' cholesterol level is quite modest.  These graphs certainly don't support the view that the lower your cholesterol, the better.

These results are observational, so they are likely subject to confounding variables and can't be used alone to determine causality.  Confounders that may affect the low cholesterol-disease relationship include the disease process (like cancer) actively lowering cholesterol and statin use to lower cholesterol in high risk people or in secondary prevention.  So some of the least healthy people may have some of the lowest cholesterol levels perhaps for no other reason than the drugs.  Confounders that may affect the high cholesterol-disease relationship include: (1) familial hypercholesterolemia which is categorised as > 240 mg/dl even through their cholesterol levels are more like 300-400 mg/dl; (2) people with ApoE4 polymorphisms because ApoE4 increases cholesterol levels but also increases oxidative stress and is a risk factor for several metabolic diseases; (3) and again, other disease states like hypothyroidism and insulin resistance that increase cholesterol, but also impair homeostasis.

Is Total Cholesterol Really a Major Factor?

The results below come from the HUNT study.  I'm showing it because unlike the three, this one has error bars and the point of the error bars is to show the variation in each group, in this case the variation in a certain cholesterol range and mortality.  As you can see there is enormous variation within each group, suggesting that there are a variety of other factors, not necessarily well correlated with cholesterol, that affect mortality.

All-Cause                    CVD                          IHD
The HUNT study.  Norwegians aged 20-74 [5]

An interesting study looked at a family tree that included people with the familial hypercholesterolemia (FH).  People with FH have a gene mutation in the LDL receptor, resulting in worse clearance and very high cholesterol (300-400 mg/dl).  They found that people with FH were likely to have lower mortality prior in the 19th century, but higher mortality in the 20th centuary.  While this isn't particularly strong evidence, it does raise some questions.  Was FH a beneficial mutation (or at least not harmful) to have prior to the 20th century?  If FH is such deleterious mutation (with regards to CVD) (as most people say it is), how come it hasn’t been strongly selected against and is still around (seeing as the increase in CVD during the 20th century and the apparent absence of CVD in hunter-gatherers can’t be explained by ‘people live longer now’) [6].   And finally, what changed in the 20th centuary to increase their mortality?  Some explanations include hydrogenated vegetable oils, food refining and smoking, all of which increased during the early 20th century.
FH and Mortality (Family Tree Study) [7]

Testing the Lipid Hypothesis

Most sources of health information, including the draft version of the 2011 Australian dietary guidelines, are written as if anything that reduces total cholesterol or LDL-C will reduce the risk of CVD (and without affecting other causes of mortality).  While this may be the case, perhaps most of the time, you can’t make that conclusion without testing.   Examples of things that lower total cholesterol but don't reduce the risk of disease/mortality include:

  • Phytosterols, which are like plant forms of cholesterol and lower cholesterol by competing with cholesterol absorption.  High dose phytosterols lower cholesterol, but don’t reduce CVD [8] [9]
  • Replacing SFA with either MUFA, PUFA or carbohydrate reduces total cholesterol, but replacing SFA with MUFA or carbohydrate doesn't reduce CVD in observational studies [10] and replacing SFA with linoleic acid (a PUFA) doesn't reduce CVD in clinical trials* [11]
  • Grain fibre reduces total cholesterol but doesn't reduce CVD in clinical trials and almost significantly increased IHD mortality (RR = 1.23 (0.97-1.57))  [12]
  • So it shouldn't be surprising that in the Women's Health Initiative (a large RCT), the prescribed low fat diet with increased fruit, vegetables and grains reduced total cholesterol, but didn't reduce mortality [13]

So please don't make the assumption that just because something that lowers cholesterol it will therefore reduce the risk of CVD, yet alone total mortality

* This is referred to as the 'Diet Heart Hypothesis' (DHH)

Further Reading:
(1) Cholesterol within nations studies
(2) The Proper Use of the Term "Lipid Hypothesis"
(3) Cholesterol and Heart Disease

Sunday, September 2, 2012


Cardiovascular Diseases and Atherosclerosis 

Coronary heart disease (CHD), also called coronary artery disease (CAD) is the narrowing of the coronary artery (the artery that supplies the heart muscle with oxygenated blood), which is often caused by atherosclerosis.  CHD may progress to ischaemic heart disease (IHD), which is a disease that involves ischemia (reduced blood supply) of the heart.  Angina is a symptom of IHD at times when the heart has insufficient blood supply. 

Rupture of unstable atherosclerotic plaque in the coronary arteries can form a blood clot (thrombus), whichs blocks blood flow to heart.  This causes a myocardial infarction (MI) (heart attack), where the heart receives an inadequate amount of oxygen to generate sufficient energy to maintain cellular homeostasis, and often some heart cells die as a result.

Atherosclerosis can also occur in the carotid artery (the artery that supplies the brain with oxygenated blood).  Then rupture of unstable atherosclerotic plaque or a blood clot may result in an ischemic stroke. 

Peripheral vascular disease (PVD), also called peripheral artery disease (PAD) can also be caused by atherosclerosis in the arteries leading to the legs, arms, etc* 

So atherosclerosis is the main cause of angina, heart attacks and ischaemic strokes.  Angina and especially heart attacks can lead to cardiomyopathy and further heart problems.  And atherosclerosis is a cause of hypertension.  While blood clots can trigger heart attacks and strokes, blood clots can be caused by ruptured atherosclerotic plaque and blood clots without any atherosclerosis is unlikely to block arteries.  I think it’s fair to say that in almost all cases: angina, heart attacks and ischaemic strokes are dependent on atherosclerosis. 

* This is probably the main reason why erectile dysfunction is so prevalent in those with CHD and may precede other signs/symptoms of CHD. 

Hypotheses of Atherosclerosis

There are a few different hypotheses as to what the initiating event in atherosclerosis is

The response-to-injury hypothesis has a few different variations.  (1) Endothelial cells wear away over time. (the cells that line the innermost part of the arteries).  But endothelial cells remain intacts in atherosclerosis and plaque build-up occurs beneath the endothelium [1].  (2) High blood pressure and inflammation cause endothelial damage, then cholesterol levels increase and the endothelial cells take up more cholesterol, where sustained injury results in a pathological accumulation of cholesterol.  Endothelial damage would increase cholesterol and cholesterol by endothelial cells [2], but only oxidised cholesterol causes atherosclerosis and cells regulate their cholesterol levels (see below).  (3) Endothelial dysfunction increases endothelial permeability, which increases the penetration and retention of lipoproteins (see response-to-retention).  Endothelial dysfunction is not sufficient to cause atherosclerosis [1], but is a contributing factor which I discuss here.

The response-to-retention hypothesis suggests the retention of lipoproteins, particularly LDL, is the initiating event in atherosclerosis.  The more LDL particles (LDL-P) the more likely they are to penetrate the endothelium.  The lipoproteins get stuck behind the endothelium by sticky proteins called proteoglycans.  This triggers an inflammatory response that oxidises the LDL particle and then macrophages engulf the stuck lipoproteins and form foam cells [1]

The oxidative modification hypothesis suggests LDL needs to be modified to initiate atherosclerosis (see below) 

Modified LDL and Atherosclerosis

Despite the labelling of LDL as ‘bad cholesterol’, regular LDL particles don’t cause the cholesterol-rich foam cells (plaques) that occur in atherosclerosis.  Even very high levels of LDL don’t lead to foam cells in vitro.  The foam cells only develop when the LDL particle is modified and there are many possible modifications (including many types of oxidative modification) [3]. 

Serum contains many antioxidants to protect against oxidative modification of LDL.  For example a 10% concentration of fetal calf serum almost completely protects against it.  It’s suggested that if LDL particles penetrate the endothelial layer they have more exposure to oxidants from smooth muscle cells and less exposure to antioxidants in the serum, which may result in LDL oxidation [3].  The problem is that LDL is the major transporter of vitamin E and atherosclerotic plaque contains a surprisingly high concentration of antioxidants, which in the case of some antioxidants (vitamin C, uric acid, vitamin E) is only slightly lower than plasma levels [1]. 

Not all oxidants are the same, there are two kinds of oxidants: 1e-oxidants and 2e-oxidants.  1e-oxidants include superoxide (O2-), nitric oxide (NO) and hydroxyl radicals (OH-), and are generally quenched by antioxidants such as vitamin C and E.  2e-oxidants include hypochlorous acid (HOCl) and peroxynitrite (ONOO-)* and antioxidants such as vitamin C and E are generally ineffective against them** [1]. 

The 1e, 2e distinction is supported by the type of LDL oxidation: 

Minimally oxidised LDL refers to LDL that is still recognised by the LDL receptor but not by macrophage scavenger receptors, and also promotes macrophage differentiation and chemoattraction.  Minimally oxidised LDL seems to be the result of vitamin E depletion and contains mostly lipid oxidation products and no protein oxidation [1]. 

Oxidised LDL (oxLDL) is where the ApoB protein is oxidised, which enables it to be recognised by macrophage scavenger receptors.  2e-oxidants preferentially react with proteins rather than lipids [1].  oxLDL has a number of additional effects such as promoting smooth muscle cell proliferation, monocyte adhesion, the production of autoantibodies, pro-inflammatory cytokines, platelet aggregation (blood clotting).  oxLDL is also toxic to endothelial cells, upregulates angiotensin receptors and production and inhibits endothelial nitric oxide synthase (eNOS) [1] [4] [5] 

oxLDL is bad stuff.  Macrophages protect endothelial cells from the toxic effects of oxLDL, and HDL particles transport oxidised lipids back to the liver.  Macrophages have a receptor for modified LDL called scavenger receptor A (SRA).  Unlike the LDL receptor, SRA doesn’t become downregulated as the cholesterol content of the cell increases [5].  What this means is that macrophages can continue to accumulate cholesterol from oxLDL but not regular LDL (or minimally oxidised LDL).

From here, macrophages that have engulfed oxLDLs become foam cells and form atherosclerotic plaque. 

* I’ll be mentioning hypochlorous acid and peroxynitrite in later posts 

** The 1e, 2e distinction may explain how pharmaceutical doses of vitamin E have pretty much failed in RCTs [1] 

Further Reading:
(1) High Cholesterol And Heart Disease — Myth or Truth?
(2) The Diet-Heart Hypothesis: Oxidized LDL, Part I
(3) The straight dope on cholesterol - Part IV
(4) Role of Oxidative Modifications in Atherosclerosis (note: it's a great paper, but only if you have time, as it's ~50,000 words long)