Fats, Cholesterol, Carbohydrates, Lipoproteins and Peroxidation: Who Are the True Atherosclerotic Assassins?
There are people in the world who are happy to spend all day scoffing vitamin- and mineral-depleted sweets but who devoutly believe that they'll instantly die of a heart attack if they eat more than three eggs per week. Eggs are like multivitamins, albeit ones that grow out of birds' multi-purpose waste vents, and yet these people treat them like deadly toxins. Why? Because they're high in cholesterol, and atherosclerotic plaques contain cholesterol, and they think that cholesterol serves no purpose other than to accumulate inside and congest the coronary arteries.

Atherosclerotic plaques also contain calcium, and hardening of the arteries is also known as "arterial calcification" - but when was the last time you saw shops selling decalcified and semi-decalcified milk?

Just because a substance - in this case calcium - is present in "clogged" arteries, it doesn't logically follow that the given substance must have caused the "clogged" arteries in the first place, or that people should righteously abstain from consuming foods that are rich in that substance. But - gosh! - that implies that, just because it happens to be present inside them, cholesterol possibly might not have caused "clogged" arteries in the first place, and that people might not be too righteous to abstain from cholesterol-rich foods.

A simple, corrective, more "enlightened" view has emerged in recent decades. According to this view, cholesterol in the diet is irrelevant because it has no effect on cholesterol in the blood, and cholesterol in the blood is caught up in the eternal struggle between good and evil. "LDL cholesterol" is "bad cholesterol" that wickedly batters and rapes the arterial wall and sets up camp inside it; "HDL cholesterol" is "good cholesterol" that comes to the relief of the beleaguered arterial walls, hacks away at the evil LDL with its trusty pickaxe, and restores a wonderful state of order and harmony in which everyone knows his/her place and zealously protects the interests of the Lord of the Manor.

According to the ditheistic worldview outlined above, "good" and "bad" cholesterol are allied to "good" and "bad" (or "healthy" and "unhealthy") fats. Saturated fat must be "bad" and "unhealthy" because it raises "bad cholesterol"; polyunsaturated fat (PUFA) must be "good" and "healthy" because it lowers "bad cholesterol"; monounsaturated fat (MUFA) must be positively saintly and godly because it lowers "bad cholesterol" and raises "good cholesterol". Most commonly consumed saturated fats also raise "good cholesterol" more than anything else does, and most commonly consumed PUFAs lower it more than other fats, but the media and the medical industry are much more interested in things that are perceived to be "bad" and "unhealthy".

Then there's "trans fat", or hydrogenated fat. Some "trans fats" - e.g. the conjugated linoleic acid (CLA) found in the meat and milk of ruminant animals - are naturally occurring, but others are created by hydrogenation at abnormally high temperatures (much higher than those encountered during regular cooking, contrary to certain claims). Hydrogenated fats, found in cheaply produced manufactured foods, must be positively "evil" because they raise "bad cholesterol" and lower "good cholesterol". Although popularly associated with saturated fats as though they are the same thing, hydrogenated fats are more often than not derived from unsaturated fats.

Then there are what must be "good bad" fats. These are the saturated fats from plant foods that can be produced and sold for handsome profits. Those from coconut oil are rare short- and medium-chained saturated fatty acids (mostly ones with 8-10 carbon chains) which, over the short term, speed up the metabolism (hence the hyperbolic promotion of them as fat-loss aids). Marketers are ever at pains to assure consumers that these wonderful fats (or "triglycerides", the name given to them so as not to alarm fat-phobes) will lower rather than raise "bad cholesterol". More on this claim in a minute. Then there are the saturated fats from cocoa butter, which are credited (along with various "anti-oxidants" found in cocoa and dark chocolate) with increasing nitric oxide (NO) production and endothelium-dependent vasodilation, and lowering blood pressure and the oxidative modification of LDL (more on the latter later). The primary saturated fats in cocoa - palmitic and stearic acid - are exactly the same ones that are predominant in beef tallow/dripping and lard, but supposedly they are magically healthy when produced by a plant rather than an animal. Luckily for cocoa and chocolate marketers, stearic acid can be converted from a "bad" 18-chained saturated fat into a saintly 18-chained monounsaturated one (oleic acid, which happens to be abundant in cocoa, tallow/dripping and lard in the first place), and 16-chained palmitic acid can be elongated into stearic acid (or desaturated into the 16-chained MUFA palmitoleic acid, which in turn can be converted into vaccenic acid, a rarer 18-chained MUFA).

Even if the "LDL=bad and HDL=good" dogma is treated as sacrosanct, the nice, neat, revised lipid-lipoprotein consensus view of "clean-eaters" and plant-fat salespersons is by no means unanimously supported by research. Although it is commonly asserted that dietary cholesterol has no effect on the levels of "LDL cholesterol" and "HDL cholesterol" in the blood, controlled feeding trials (those in which the subjects eat under strict supervision or at least are provided with meals that differ only in the content of certain macronutrients and whatever micronutrients happen to occur alongside them) show a small overall raise in both, this being the mid-point between "hyperresponders" (in whom dietary cholesterol significantly raises "LDL cholesterol" and "HDL cholesterol") and "hyporesponders" (in whom it doesn't): 1992 Meta-Analysis. In a not-strictly-controlled feeding trial of Mexican schoolchildren, supplementing with two whole eggs (518mg of cholesterol) per day increased both "LDL cholesterol" and "HDL cholesterol" in "hyperresponders" and had no effect in "hyporesponders", and resulted in larger LDL particle sizes in all subjects and a shift from LDL phenotype B to LDL phenotype A (more on LDL particle sizes and phenotypes later).

Short- and medium-chained saturated fatty acids, those with 10 carbon chains or fewer, are fervently believed to lower "LDL cholesterol" levels, and are most prevalent in coconut oil, but coconut oil's most abundant fatty acid, 12-chained lauric acid, has been democratically elected by meta-analysis as the most potent raiser of "LDL cholesterol" (and the most potent raiser of "HDL cholesterol" and the ratio of "HDL cholesterol" to "LDL cholesterol"): 2003 Meta-Analysis. Accordingly, dietary supplementation with coconut oil increased "LDL cholesterol" levels while beef fat left them unaltered when the two were compared in one feeding experiment (beef fat lowered "HDL cholesterol" and raised triglycerides). In a strictly controlled in-patient feeding trial and a not quite so strict one, adding medium-chain saturated fatty acids (approximately two-thirds 8-chained caprylic acid to one-third 10-chained capric acid) resulted in significant increases in "LDL cholesterol" (and its ratio to "HDL cholesterol") compared to oleic acid in both, in significant or borderline-significant increases in triglycerides and VLDLs (more on these later) compared to oleic acid in both and compared to palm oil in the first, and in significant hepatic de novo synthesis of palmitic and palmitoleic acid compared to oleic acid in both and compared to palm oil in the first (see below for more on the potential significance of this). Only oleic acid was compared to the medium-chained saturated fatty acids in the second experiment.

As for "bad" saturated fats being converted into "good" monounsaturated ones, as in stearic acid transforming into oleic acid, it should be noted that this conversion does not automatically bring about the popularly attributed effect of saturated and monounsaturated fats on levels of "LDL cholesterol" and "HDL cholesterol". Different sub-fractions of saturated and monounsaturated fats have their own independent effects. Stearic acid has no effect on "LDL cholesterol" (or triglycerides) and increases "HDL cholesterol", and it appears that palmitoleic acid, the 16-chained MUFA derivative of saturated palmitic acid, behaves in the same fashion as its precursor by slightly raising "LDL cholesterol" - but, unlike palmitic acid, it lowers rather than raises "HDL cholesterol".

You may be wondering what the standard of reference is when people talk of different fats raising or lowering substances with supposedly contrasting moral fibres. That standard of reference, in the 1992 and 2003 meta-analyses cited above, was carbohydrate, the macronutrient which mainstream nutritionists and dieticians insist should contribute 50-60% of our diet. "Healthy" carbohydrate raises "bad cholesterol" compared to most MUFAs and PUFAs, and lowers "good cholesterol" compared to most saturated fats, MUFAs and even PUFAs. It proportionally raises triglycerides as it lowers "good cholesterol". It increases the ratio of "bad cholesterol" to "good cholesterol" compared to all fatty acids except for leaving it about the same as myristic acid (the 14-chained saturated fatty acid) and palmitic acid (and presumably palmitoleic, caprylic and capric acid as well). Two out of three may not be bad, but one out of three seems a little bit crap, and zero out of three plainly sucks. Luckily, logic is not high among the priorities of mainstream nutrition.

Polyunsaturated fats are sometimes referred to as "the essential fatty acids". This is because the body cannot create them from other substances, and therefore it is essential to obtain them in the diet. Saturated fat is technically non-essential in the diet because the body can convert glucose (derived from dietary carbohydrate) into various saturated fatty acids. Humans who eat a low-fat, high-carbohydrate diet display higher quantities of two supposedly "atherogenic" saturated fatty acids in their red cells, plasma phospholipids and cholesterol esters. Which two? Yes, you guessed it: glucose derived from dietary carbohydrates can be converted via hepatic de novo lipogenesis into myristic and palmitic acids (the only two saturated fats that don't increase the ratio of "good cholesterol" to "bad cholesterol" compared to carbohydrate - possibly excepting caprylic and capric acids, which also lead to hepatic de novo lipogenesis)! Palmitic acid derived from hepatic lipogenesis can be further converted into palmitoleic acid, which increases the ratio of "bad cholesterol" to "good cholesterol" even more so! Monounsaturated fat is technically non-essential in the diet because certain saturated fatty acids (see above) can be converted into monounsaturated ones. So, although not strictly essential in the diet, saturated fats and MUFAs are essential in the body - and are present in much higher quantities than "essential" PUFAs ...

Although stearic acid can be converted into oleic acid, oleic acid cannot be converted into the "essential" 18-chained PUFAs. PUFAs are divided into omega-3 and omega-6 classes, each with a variety of fatty acids with different numbers of carbon chains. The "essential" 18-chain PUFAs are alpha-linolenic acid (omega-3) and linoleic acid (omega-6). There is also an 18-chain "omega-9" PUFA, mead acid, and oleic acid can be converted into this, but this is a useless, non-essential substance. It is not clear whether alpha-linolenic acid (ALA) and linoleic acid (LA) are essential in their own right, but ALA can be converted into docosahexaenoic acid (DHA, omega-3, 22 carbon chains) and LA can be converted into arachidonic acid (AA, omega-6, 20 carbon chains). DHA and AA are essential for proper brain function, but the rate of conversion from their 18-chained precursors is crap. Both can be obtained directly from eggs laid by grass- and insect-fed hens; AA can be obtained directly from almost all animal fats (not dairy ones); DHA can be obtained directly from mammal brains, fish, shellfish and non-terrestrial molluscs.

Small doses of DHA-containing fish oil have led to beneficial health outcomes in some trials, including those concerned with coronary heart disease (CHD), but not in all (pdf). This superficially fits in nicely with the idea that PUFAs (especially omega-3s) are "good" and "heart-healthy" things that lower "bad, artery clogging LDL cholesterol". One tiny problem: the research showing that PUFAs lower "bad cholesterol" is overwhelmingly based on the feeding of vegetable oils rich in omega-6 linoleic acid. The authors of the above-cited 1992 and 2003 meta-analyses acknowledged that they had excluded omega-3 fatty acids from consideration because of evidence showing that those from fish oil (but not alpha-linolenic acid) raise "LDL cholesterol".

Fish oil, the fat from fish, typically contains fairly significant portions of all three major classes of fatty acids. If it is derived from the liver, it also contains super-high concentrations of fat-soluble vitamins, particularly A (retinol) and D (cholecalciferol). You may note that the name for vitamin D (more specifically vitamin D3) - cholecalciferol - sounds suspiciously like cholesterol (which, incidentally, is also present in untampered-with fish oil). Cholesterol is the precursor to cholecalciferol, and to "steroid" hormones such as testosterone, and therefore serves many purposes besides supposedly clogging arteries. Also present in fish oil may be other fat-soluble vitamins, E and K2. Different fish oils and products contain different properties depending on where they've come from and what's been done to them.

The longer-chained omega-3s found in fish oil, DHA and eicosapentaenoic acid (EPA, 20 carbon chains), do indeed have a very mixed record where "bad cholesterol" is concerned. There have been odd occasions on which they have lowered it (in this case higher-DHA tuna and salmon were responsible while higher-EPA pollock had little effect). Usually, however, they have had no effect or have raised it - and, despite the results of the study just cited, DHA has often been responsible for the raising:


?Fish oil of unspecified composition raised "LDL cholesterol" and increased the size of LDL particles, and serum extracted from those fed it oxidized more readily
?Mixed EPA-DHA fish oil decreased "total cholesterol" whether added to a diet higher or lower in saturated fat - but increased the ratio of "LDL cholesterol" to "HDL cholesterol"!
?"LDL cholesterol" increased and oxidation of LDL decreased in response to a supplement with a 2-1 ratio of DHA-EPA
?Purified DHA but not purified EPA increased "LDL cholesterol" and decreased "small, dense LDL"
?DHA supplement increased "LDL cholesterol" and apolipoprotein B - and the Ratio of the former to the latter, indicative of less "small, dense LDL"
?DHA supplement increased "LDL cholesterol" and reduced "small, dense LDL"
?DHA-supplemented eggs increased "LDL cholesterol" and "HDL cholesterol", and decreased triglycerides and "small, dense LDL" (abstract)
?DHA raised "LDL cholesterol" not quite significantly and lowered triglycerides and "small, dense LDL"
?Platelet phospholipid DHA content correlated with "LDL cholesterol" increases, EPA content with triglycerides decreases, and both with increased oxidation of extracted LDL

On one amusing occasion (abstract), fish oil or fatty fish consumption reduced the rate of mortality from any cause while also increasing the level of "bad cholesterol". Failure to sufficiently reduce "LDL cholesterol" was blamed on the absence of any effect from a low-fat-but-high-PUFA diet, but the full text, not available online, reveals a raised "LDL cholesterol" level in the fish-eaters enjoying greater longevity!
This brings me to two minor problems with the notion of "bad cholesterol":


1.It's not bad (at least not in all of its forms)
2.It's not even cholesterol (at least not as many people conceive it)

Other than that, I embrace the concept wholeheartedly. Cholesterol is not fat but, like fat, it is a "lipid" and not soluble in water. Protein is water-soluble, and fats and cholesterol therefore hitchhike rides around the body's water-rich blood vessels in entities called "lipoproteins". LDL stands for "low-density lipoprotein"; HDL stands for "high-density lipoprotein". There are other fat- and cholesterol-transporting lipoproteins, too. Chylomicrons are produced in the gut and are broken down by the liver into "very-low-density lipoproteins" (VLDL). VLDLs are rich in "triglycerides" (or "triacylglycerols" or "triacylglycerides"), but all lipoproteins contain some triglycerides (three fat molecules plus one molecule of glucose-derived glycerol). VLDLs are gradually worn down into "intermediate-density lipoproteins" (IDL, fairly high in triglycerides) and then into LDLs. Not all LDLs are identical, either. Some are smaller and denser and less cholesterol-rich than others (small, dense LDL or LDL subtype/phenotype/pattern B, as opposed to large, buoyant, cholesterol-rich LDL or LDL subtype/phenotype/pattern A), and some combine with something called "apolipoprotein a" [apo-(a)] to make "lipoprotein a" [lp(a)]. Apolipoproteins are lipoprotein derivatives, the most commonly analysed ones being apo-(a) and apo-(b), the latter present in all VLDLs, IDLs, small and large LDLs and lp(a)s. LDLs (often called "bad cholesterol") and HDLs (often called "good cholesterol") are not cholesterol, and the cholesterol that they do carry is identical. Sometimes, however, LDL can become "oxidized". Cholesterol can also become oxidized ("oxysterol"), but this is not the same thing as an oxidised lipoprotein.

Are you sufficiently confused yet? No? Then you might like to bear in mind that the "blood/serum cholesterol" measurements usually taken do refer specifically to the cholesterol in LDL and HDL particles (which is plain old cholesterol regardless of the particle in which it's carried), but that "oxidized LDL", "small, dense LDL" and "large, buoyant LDL" refer to different varieties of low-density lipoprotein (which are not cholesterol but which do carry some). Factors that increase the amount of cholesterol inside a lipoprotein fraction do not necessarily cause an increase in the number of those lipoprotein particles.

In most cases, those examining the relationship between cholesterol-inside-lipoprotein levels and mortality from specific or general causes have not distinguished between different varieties of LDL. Masses and masses of research has been carried out concerning the strong correlations (or the embarrassing lack of them) between undifferentiated "LDL cholesterol" and the presence of early death and disease from CHD and all causes among populations of different ages and sexes. I shall focus only on research distinguishing different forms of LDL.

High fasting levels of small, dense LDL (subtype/phenotype/pattern B), combined with high fasting levels of insulin and apo-(b), predicted the presence of ischaemic heart disease (IHD) in non-diabetics, and making statistical adjustments for this triumvirate meant that the ratio of LDL-HDL was no longer predictive of IHD (and neither were triglycerides). The fasting levels were taken only at the start of the 5-year follow-up period - a common drawback in studies of this sort.

A 13-year follow-up of a different population, again going on baseline values, found that small, dense LDL strongly and independently predicted the development of IHD, and that large, buoyant LDL negatively predicted it. The latter association held good over the 13 years, but the link between small, dense LDL and IHD was only very remarkable over the first 7 years.

A 6-year follow-up of another population revealed that baseline "non-HDL" (all lipoproteins except HDL) was a much stronger predictor of coronary heart disease (CHD) than LDL alone, and mutual adjustment between "non-HDL" and LDL meant that the latter was no longer predictive at all. Apo-(b) by itself was a stronger predictor than "non-HDL", and "non-HDL" was no longer predictive once that and apo-(b) were adjusted for one another. Apo-(b) is present in all the potential "culprits", including lipoprotein(a), small and dense LDL and oxidized LDL, the last two of which were not measured here.

Lipoprotein(a) bore no interesting relationship to CHD in the above analysis, but it is often a better predictor of cardiovascular mortality than plain "LDL cholesterol":


?Lp(a), Apo-(b) and Apo-(b) versus Apo-(a) Ratio (abstract)
?Lp(a) and CAD, Peripheral Vascular Disease, Ischaemic Stroke and Abdominal Aortic Aneurysm
?Lp(a) and CHD Meta-Analysis

Lipoprotein (a) concentrations tend to correlate very strongly with those of oxidized LDL. A high level of the former was the only factor from which a high level of the latter was not independent in predicting the presence of coronary artery disease (CAD) among over 500 subjects about to undergo coronary angiography (oxidized LDL was predictive independent of all other factors among those under 60).

In the following study, levels of cholesterol in LDL did not differ (and were even below the supposed desirable upper limit) among patients with acute myocardial infarction (AMI, the posh for "heart attack") and unstable or stable angina pectoris (posh for "chest pain"), but oxidized LDL was very significantly higher in those with AMI, and those with UAP had much higher levels than controls. Atherectomies from different patients with unstable angina pectoris (UAP) bore much greater evidence of oxidized LDL in macrophage-derived foam cells compared to those from patients with stable angina pectoris (SAP).

UAP is divided into three different "Braunwald" classes of extremity, and those in the highest class had much higher levels of oxidized LDL (pdf). Being in Braunwald Class III independently predicted the presence of complex lesions, but some in the lower classes also had complex lesions, and these patients typically had elevated levels of oxidized LDL. Oxidized LDL was by far the strongest predictor of complex lesions. Levels of cholesterol in LDL didn't differ between the different groups, again being close to the supposed desirable upper limit.

Arterial or coronary calcification is a common feature in those with SAP. In the following study, those with AMI had lower "total coronary calcium scores" than those with SAP - but far higher levels of oxidized LDL (pdf). Again, levels of cholesterol in LDL were the same and a little below the supposed desirable upper limit.

In the following study, on aortic stenosis patients whose aortic valves had been replaced, the levels of cholesterol in LDL did not differ but levels of small, dense LDL (and triglycerides) were significantly higher in those whose explanted valves contained the most oxidized LDL. Oxidized LDL and small, dense LDL are frequently elevated together because, as outlined in the following highly detailed abstract, small, dense LDL does not readily interact with the LDL receptor, spends longer in the blood, and oxidizes and enters the arterial wall more readily. Oxidized LDL and small, dense LDL occur alongside elevated triglycerides and VLDL, reduced HDL and greater accumulations of visceral/intra-abdominal fat (as distinct from subcutaneous fat) in the "Metabolic Syndrome" or "Syndrome X".

So, levels of triglycerides and VLDL (raised by carbohydrate) might be better predictors of disease and premature death than boring old "LDL cholesterol" (raised by saturated fats with 8-16 carbon chains, monounsaturated palmitoleic acid and DHA). What about small, dense LDL, lipoprotein(a) and oxidized LDL? They seem to be better predictors than "LDL cholesterol", but what dietary factors cause them to be increased?

When male subjects over 20 were fed a 6-week out-patient diet containing 24% fat and 60% carbohydrate rather than 46% fat and 38% carbohydrate (the ratio of saturated fat to PUFA being kept constant), their "LDL cholesterol" levels fell due not to a reduction in particle numbers but due to an increase in small, dense, cholesterol-depleted LDL particles at the expense of large, buoyant, cholesterol-rich ones! The effect was more pronounced in subjects possessing apolipoprotein E [apo-(e)] 4/3 and 4/4 allele phenotypes.

One third of a similar group of subjects in a similar earlier trial converted from LDL phenotype A (large, buoyant) to phenotype B (small, dense) as a result of following a low-fat, high-carbohydrate diet. The "remedy" was to inflict a 10% fat, 75% carbohydrate diet on 38 men who had stubbornly sustained phenotype A on 20-24% fat. The result was that a further 12 subjects got to savour the joys of a small, dense LDL phenotype B. 

When 58 males and 26 females with an average age in the early 20s were fed a higher-MUFA (22%) and then a higher-carbohydrate (55%) diet (or vice versa) for 4 weeks each following an opening 4-week higher-saturate diet (20%), LDL became smaller, denser and more phenotype-Ber on the high-carbohydrate regimen - except in those sporting the apo-(e) 4/3 allele phenotype (contrary to the above). [Having looked at these "contradictory" studies in closer detail in a section of my article on types of carbohydrate and lipoprotein fractions et al, they no longer seem so contradictory.] Those with the 4/3 phenotype had significantly larger, buoyanter LDL particles after the high-carbohydrate diet compared to the high-MUFA diet - though their particles were larger and buoyanter after all the diets compared to those with other apo-(e) phenotypes. The PUFA content of all the diets was 6%, and the carbohydrate content of the "high-fat" diets was 47% (compared to 38% fat).

Saturated fat and MUFA were not seen to differentially affect LDL particle size in the above trial. In another trial, in which 30 males and 26 females with an average age in the mid-20s were for 4 weeks fed under boarding school conditions "high-fat" diets (35-40% fat, 45-50% carbohydrate) that were identical except for being supplemented by olive oil (high-MUFA), rapeseed/canola oil (highish-MUFA with a not insignificant PUFA content from a mix of omega-6 linoleic acid and omega-3 alpha-linolenic acid) or sunflower oil (high-PUFA from linoleic acid), all of the diets resulted in a small but significant reduction in the size of LDL particles compared to a 2-week baseline diet that was higher in saturated fat. The source of the saturated fat in the baseline diet is not mentioned. The authors, who were supported by grants from, among other bodies, the German Union for the Promotion of Oil- and Protein-Containing Plants, emphasised that the reductions were very small - but the statistical significance (0.01) was greater than the 0.05 that usually has authors wetting themselves in excitement.

Scanning through the earlier links concerning the effect of long-chained omega-3 fats on "LDL cholesterol" levels should make it clear that adding DHA to a diet often results in larger, more buoyant, more cholesterol-rich LDL particles - even though it sometimes increases "in vitro" oxidative susceptibility.

When 29 men and 29 women aged 25-65 spent 6 weeks eating (mostly under supervision) four different 39-40%-fat, 45-46%-carbohydrate diets, high in the LDL-cholesterol-raising saturated fatty acids (lauric, myristic, palmitic), high in oleic MUFA, moderate in naturally occurring trans fatty acids and partially hydrogenated ones or high in both of these, lipoprotein(a) levels lowered on the diet high in the 12-to-16-chained saturated fats but remained unchanged on the others. 

What effect does an ultra-virtuous low-saturate, 8-portions-of-grain-per-day, sometime fruit-and-vegetable-enriched diet have on lipoprotein(a) and oxidized LDL? 37 females spent 2 5-week periods (with a 3-week washout period) eating at a hospital kitchen diets fitting the above description, one higher (c. 10 portions per day) and another lower (2 portions per day) in berries, fruits and vegetables. Both experimental diets contained about 30% fat (c. 10% saturated) and 50% carbohydrate. Result? Oxidized LDL and lipoprotein(a) increased significantly in response to both experimental diets. 

In another experiment supported by a grant from the redoubtable German Union for the Promotion of Oil- and Protein-Containing Plants, 31 men and 27 women in their mid-20s spent 4 weeks being fed (under boarding school conditions) diets that were near-identical except for being enriched with olive oil, rapeseed oil or sunflower oil (see 4 paragraphs above for the characteristics of these oils). The olive oil diet increased the lag time before extracted LDL oxidized "in vitro", and decreased the conjugated diene propagation rate and the maximum amount of conjugated dienes, compared to sunflower oil. Rapeseed oil was always somewhere between the others. Likewise, the LDL alpha-tocopherol content was always higher on the olive oil diet and lower on the sunflower oil diet (despite the even higher alpha-tocopherol content of the latter), with rapeseed (very high in gamma-tocopherol) intermediate. The 4-week experimental diets were preceded by a 2-week diet higher in saturated fat. The LDL extracted from the subjects in this phase was less susceptible to oxidation only in comparison to that from the sunflower oil diet, and was more susceptible than that from the olive oil and rapeseed oil diets. Unfortunately, the higher-saturate diet (of unspecified source) was not near-identical to the others in every respect, containing 2 to 3 times less vitamin E from alpha-tocopherol. The authors downplayed the importance of alpha-tocopherol or other tocopherols, noting that other tocopherols are not incorporated into LDLs in large quantities, but didn't mention that other tocopherols (especially gamma-) and a variety of other properties could spare LDL and its constituents from oxidation (more on this in a while).

A different experiment compared the effects of different fats (butter, palm oil, olive oil, rapeseed oil, sunflower oil) on postprandial oxidation of VLDLs and LDLs. Not surprisingly, no effect was seen on the latter, which derive from the former and do not have time to form to any appreciable extent in the postprandial period. 18 males aged 22-35 on 6 widely separated occasions ate meals from which the postprandial VLDLs were extracted to be subjected to oxidation. The results showed that factors other than the "unsaturation index" also affect the oxidative susceptibility of lipoproteins. Olive, rapeseed and sunflower oil relatively affected the oxidation lag time and conjugated diene propagation rate as they did in the experiment above. Interestingly, butter resulted in the highest ratio of oleic to linoleic acid in VLDLs, and had the lowest "unsaturation index" (a calculation based on the quantity and types of PUFAs), but increased lag time to oxidation less than palm oil (39% saturated, 47% MUFA, 14% PUFA, but with higher quantities of alpha- and gamma-tocopherol and probably tocotrienol forms of vitamin E and certain other "anti-oxidants"). Butter resulted in the lowest postprandial propagation rate, but it was not significantly lower than that of palm or olive oil. Sunflower and rapeseed oil resulted in postprandial propagation rates significantly higher than those of all the others, despite their higher tocopherol content. Thus neither degree of saturation nor alpha-tocopherol can fully explain a fat's effect on LDL oxidation. Another point worth noting, as noted by the authors of this experiment, is that "in vitro" measurements of LDL oxidation (i.e. the oxidative damage that LDL sustains after having been removed from the blood and its protective water-soluble nutrients including vitamin C) don't necessarily reflect real life.

Cocoa beans, cocoa mass, cocoa powder and cocoa-rich chocolate are rich in polyphenol flavonoid "anti-oxidants" that are credited with the frequently observed effect of these products reducing the oxidation of LDL, among other things. Cocoa butter (the fat-containing component of whole cocoa products) contains little alpha-tocopherol but ought to be reasonably high in gamma-tocopherol (the USDA Nutrient Database ascribes almost 6mg of gamma-tocopherol to a 100g portion of unsweetened baking chocolate containing 50-55g of fat). The gamma-tocopherol could well play a role in reducing the oxidation of LDL. Nevertheless, daily feeding of 13, 19.5 or 26g low-fat (and therefore low in gamma-tocopherol) sucrose-enriched high-polyphenol cocoa beverages to 69 male and 91 female adults of diverse ages for 4 weeks reduced oxidized LDL in those given the polyphenol-containing beverages compared to those given a control beverage which contained cocoa butter and which mimicked the others except for having only the faintest trace of polyphenols. It was not a strictly controlled feeding trial, although the subjects were provided with their dinners, and their overall diets (judged by 3-day food records) seem remarkably similar. It's therefore hard to attribute the effect to anything other than the polyphenols (all beverages contained 3.7g of cocoa butter and 12g of sucrose). That said, the effect was not dose-responsive, indicating that, if polyphenols are heroes, they are so in fairly modest quantities. They function as anti-nutrients as well as anti-oxidants, but that's another story.

On another, rather amusing occasion, supplementing subjects for 3 weeks with 75g per day of white chocolate (probably rich in gamma-tocopherol) or high- or low-polyphenol dark chocolate reduced LDL conjugated dienes equally in all groups (abstract)!

However, one of the polyphenol flavonoids from cocoa, epicatechin, has been shown with near-certainty to be responsible (partly or fully) for the effect of cocoa on nitric oxide production and flow-mediated vasodilation. A high-flavonoid cocoa increased these more than a low-flavonoid one, and circulating concentrations of epicatechin and one of its metabolites predicted the effect, which was also produced by oral administration of pure epicatechin (and prevented by inhibition of "nitric oxide synthase", NOS). It should be noted that a lot of research carried out by this team was funded by Mars (the chocolate manufacturers, not planet or the Greco-Roman God of War).

So, factors other than the degree of fat unsaturation affect the oxidative susceptibility of LDL, but there is no doubt that PUFA-dominant oils work to increase "in vitro" LDL oxidation. A Food Pyramid or "healthy eating" diet increases oxidized LDL and apolipoprotein(a). High-carbohydrate diets usually make LDL particles smaller and denser compared to high-fat diets, and high-MUFA and high-PUFA diets seem to do the same compared to high-saturate diets. All of these things have valid claims to being better predictors of heart-related mortality than the amount of cholesterol contained in LDL particles. If saturated fat did any of these things, it would be immediately accepted by most people as unquestionable proof of its supreme evilness. But it doesn't do any of these things, and all we ever hear about is boring old "good" and "bad" cholesterol-transporting lipoproteins - which may actually be severely depleted of cholesterol if the subjects have been adhering piously to a nanny-state nutrition "healthy eating" regimen!

Taking the above into account, the most worrisome macronutrients appear to be carbohydrate and PUFA. Does this mean that they are intrinsically "unhealthy" and that, though they may have uses in small doses, they should only be eaten in strict "moderation"? This seems nice and straightforward, but skepticism should always be up-regulated when something seems excessively nice and straightforward.

A high alpha-tocopherol content doesn't protect sunflower oil from oxidative damage, but highly polyunsaturated oil that has been extracted from its original context in a food will obviously be exposed to more air and therefore more oxidative damage. Walnuts are very high in PUFA. Have you ever noticed the difference in taste between those freshly broken out of the shell and those that have already been removed and packed into a bag? The former are usually very pleasant (provided not too much of the shell structure is ingested accidentally); the latter are utterly vile (they may taste nice for a moment, but the taste quickly alters and you have visions of toxic fungi changing colour).

In spite of my distaste for those pre-removed from the shell, it's only fair to point out that a not strictly controlled feeding trial generously funded by the California Walnut Commission found that eating ready-bagged walnuts (rather than olive oil) did not increase the oxidative susceptibility of LDL. Another trial funded by the same commission found much the same thing - although one of the measures of LDL oxidative susceptibility was significantly increased (see table 3). In yet another trial funded by the same commission, adding walnuts to subjects' habitual diets significantly increased large, buoyant LDL and decreased small, dense LDL - a trend that reversed during a subsequent lower-fat diet. In a stricter feeding trial, for which the walnuts were kindly donated by the California Walnut Commission (which also provided funds), eating walnuts in replacement of other sources of fat from the Japanese diet did not increase the oxidative susceptibility of LDL extracted from the women (apparently they didn't bother to extract it from the men).

PUFAs from fish are also likely to vary in oxidative susceptibility depending on how fresh they are. Looking at the studies cited earlier makes it clear that their record on the oxidative front is not entirely consistent. Taste is probably as good a guide as any. Fresh sockeye salmon is delicious (with butter); that from tins is foul (regardless of what you add to it). Even "fresh" supermarket mackerel is appallingly rancid, but ultra-fresh mackerel bought by the sea is, if not tantalisingly tasty, at least suggestive of good nourishment.

Not unlike rapeseed oil, almond fat is predominantly rich in MUFA but also has a fairly significant PUFA (linoleic acid) content (almost as much linoleic acid as rapeseed but with little or no alpha-linolenic acid and therefore an "evil", "evolutionarily incorrect" high ratio of omega-6 to omega-3). Eating 73g of whole, unblanched almonds per day (in preference to a whole-wheat muffin that was ultra-low in saturated fat) greatly reduced oxidized LDL - albeit in the context of a "therapeutic" low-fat National Cholesterol Education Program Step 2 diet (with funds provided by the Almond Board of California). On another occasion, again with funds from the Almond Board of California, adding fat to the diet via whole almonds or almond oil (in replacement of saturated fat) had no effect on LDL oxidation. From this limited evidence it appears that, whether it is left within the seed or extracted as oil, almond fat (high-MUFA but also 15-20% PUFA from linoleic acid) has no effect on LDL oxidation when replacing saturated fat, and greatly reduces LDL oxidation when replacing NCEP-approved low-saturate whole-wheat muffins.

What of carbohydrates? Unlike fat, carbohydrate cannot be stored by the body in large amounts. It can be stored as glycogen in the muscles and organs, but 4 days of a very-low-calorie diet (405kcal) are sufficient to deplete glycogen stores in regular subjects. Potassium is stored with glycogen, and the total potassium losses over the first 4 days of the just-cited study were nearly twice as great the total losses over the subsequent 66 days of the same diet. Around 400g was the average estimated initial glycogen store of the 11 female subjects, though the range was from 80g (for a 107.6kg subject) to 1066g (for an 84.4kg subject). Supposing a male athlete weighing the same as the latter subject stored 2000g of glycogen, that would represent only 2kg of weight and the hypothetical athlete would still have more weight stored as fat if his body-fat percentage was as low as 2.5.

"Western" populations contain many individuals whose glycolytic (glycogen-utilising) energy expenditure and glycogen turnover are very low, but who think it judicious to imitate the feeding strategies of Tour de France competitors. Result? Lots of time with high levels of carbohydrate-derived glucose in the blood. Humans, most primates and a few other animals do not possess the enzyme that converts glucose into vitamin C (the memorably titled L-gulono-gamma-lactone oxidase). Consuming lots of vitamin C temporarily allows higher than normal blood glucose levels and might negate or reduce the damage done by such levels (scroll to page 1218 of this pdf to read about a case of 4500mg per day of vitamin C having the "adverse" effect of making doctors look silly by misdiagnosing a man as diabetic), but few people consume such large quantities of vitamin C.

It would be naive to assume that there is a simple, dose-responsive relationship between carbohydrate consumption, blood glucose levels, "atherogenic" lipoprotein fractions and the presence of disease. Too high or too low a carbohydrate consumption could be damaging, but "high" and "low" should not be taken to mean fixed quantities or percentages arbitrarily ordained by cereal fetishists on one side and carbo-hydro-phobes on the other. They should be judged in accordance with demand and turnover, which in many "westerners" happen to be significantly lower than the usual supply.

Here's a summary of the main points of this article:

?LDL (often called "bad cholesterol") and HDL (often called "good cholesterol") are not cholesterol but they are transporters of it (lipoproteins)
?LDL (low-density lipoprotein) can be larger, more buoyant and cholesterol-rich (pattern A) or smaller, more dense and cholesterol-depleted (pattern B), and only the latter appears to be "bad"
?HDL (high-density lipoprotein) also comes in different shapes and sizes, but few people care about them, and I'm not currently one of them
?LDL can become oxidized and it appears to be much "badder" in this form
?LDL combines with "apolipoprotein a" to form lipoprotein(a), and the latter has a "bad" stigma
?LDL derives from VLDL (very-low-density lipoprotein) via IDL (intermediate-density lipoprotein), and VLDL has a "bad" stigma
?Nobody seems to care about IDL, and I'm not sure that I do either
?VLDL is rich in triglycerides/triacylglycerol/triacylglycerides, but all lipoproteins contain some, and triglycerides have a "bad" stigma
?"Apolipoprotein b" is present in all of the lipoproteins that have a "bad" stigma, and consequently has a "bad" stigma of its own
?Dietary cholesterol increases the amount of cholesterol in HDL and large, buoyant LDL rather than in small, dense LDL
?Saturated, monounsaturated (MUFA) and polyunsaturated (PUFA) fats are not singular entities but all occur in a great variety of forms
?Carbohydrates also occur in a great variety of forms, even though they're often treated as a singular entity in regard to lipoprotein fractions, including by me in this article (maybe I should write another)
?"Unhealthy" saturated fats with 12-16 carbon chains increase "good cholesterol" and lower lipoprotein(a), and one of them (12-chained lauric acid) increases the ratio of "good cholesterol" to "bad cholesterol" more than any other fatty acid
?"Healthy" saturated fats with 8-10 carbon chains increase the ratio of "bad cholesterol" to "good cholesterol", increase VLDL and triglycerides and are converted into "unhealthy" 16-chained saturated palmitic acid via hepatic de novo lipogenesis
?"Healthy" carbohydrate increases small, dense LDL, lowers "good cholesterol" and, like "healthy" medium-chained saturated fats, increases VLDL and triglycerides and can be converted via hepatic de novo lipogenesis into "unhealthy" saturated fats (14-chained myristic acid and 16-chained palmitic acid)
?Palmitic acid (as produced via hepatic de novo lipogenesis from "healthy" carbohydrate or medium-chained saturated fats) can be desaturated into a 16-chained monounsaturated fat (palmitoleic acid) that raises "bad cholesterol" and lowers "good cholesterol"
?"Healthy" unsaturated fats increase small, dense LDL compared to saturated fats of unspecified type
?Those saturated fats of unspecified type failed (in the context of a diet much lower in alpha-tocopherol) to protect extracted LDL from "in vitro" oxidation
?"Healthy" omega-6 polyunsaturated fat lowers "good cholesterol" compared to other fats, and increases the "in vitro" oxidation of extracted LDL
?"Healthy" omega-3 polyunsaturated DHA increases "bad cholesterol", increases large, buoyant LDL at the expense of small, dense LDL, and sometimes increases the "in vitro" susceptibility of LDL to oxidation
?"Healthy" omega-3 polyunsaturated EPA decreases triglycerides and sometimes increases the "in vitro" susceptibility of LDL to oxidation
?Everything is more complicated than it seems