Part I - DOES CHOLESTEROL CAUSE ATHEROSCLEROSIS AND CORONARY HEART DISEASE?

INTRODUCTION
Cholesterol causes atherosclerosis and coronary heart disease, or put another but not equivalent way, there is a positive association between cholesterol levels and development and extent of atherosclerosis and thus coronary heart disease. Everyone knows this. This hypothesis has the status of an unquestionable belief, a self-evident truth. It is important, however, to distinguish between atherosclerosis and symptomatic or actual coronary heart disease (CHD). The latter generally includes angina or a history of one or more fatal or non-fatal heart attacks. Thus when CHD is an endpoint in a study, this generally includes clinical manifestations and actual adverse events, whereas if atherosclerosis is the subject of study, merely its presence and extent are at issue. Nevertheless, there is of course a close connection between the two. If an individual has no coronary atherosclerosis at all, the risk of an adverse coronary event is very small, and risk of symptomatic CHD or CHD adverse events increases with the extent of atherosclerosis. Also, the observation of atherosclerotic deposits in the coronary arteries leads to the diagnosis of coronary heart disease which makes itself evident by both symptoms (angina) and events (heart attacks). However, if one restricts their attention to the heart attack event itself, then the situation is somewhat more complex, given what in general is a multi-step process.

One might have thought that the cholesterol hypothesis would have met with resistance. After all, it seems a bit curious that a substance alleged to cause atherosclerosis and thus heart disease in fact comprises 5% of cell membrane lipids and plays a key role in maintaining cell wall structure. It is also the starting point for the synthesis of several groups of very important biochemicals, including male and female sex hormones, vitamin D (via photochemical action in the skin) and bile acids. Cholesterol is also used by the body in natural healing processes and tissue repair. The absorption through the gut of dietary cholesterol is poor, and the body is generally able to compensate for dietary intake by adjusting the endogenous synthesis which occurs mainly in the liver, and thus for most individuals, the serum levels are only very weakly related to dietary intake. Also, cholesterol molecules are never found free in blood since they are insoluble and are transported mainly by so-called lipoproteins of which the high and low density varieties (LDL and HDL) are regarded as bad and good according to the conventional wisdom, in spite of the fact that transport of cholesterol is essential to life. Given the vital functions of this molecule and its lipoprotein transporters, is it not a bit surprising that cholesterol is a cause of atherosclerosis?

Normal progression in science involves falsifying hypotheses rather than proving them. When there are associated with a hypothesis a number of inconsistencies, observations that appear to contradict or falsify the hypothesis, then there should be cause for concern that it is flawed or false and might in fact be leading down a dead-end road. The Bohr atom studied by every high school chemistry student is an example where a beautiful hypothesis (theory) was eventually falsified and replaced by quantum theory. But Einstein spent much time trying to find a so-called thought experiment to demonstrate that one of the essential pillars of the quantum theory was false. Incidentally, he failed but this is how progress is made. But today, anyone who questions the hypothesis that cholesterol in general or LDL cholesterol in particular causes or is a risk factor for atherosclerosis is ignored, branded a nut, or ostracized by professional colleagues, or perhaps told not to worry because it is really oxidized LDL that is important—just wait until this revision becomes a genuine truth that no one can question. Cholesterol is really LDL is really oxidized LDL, no problem. In the literature it is repeatedly stated that the evidence backing the Cholesterol Hypothesis is overwhelming, but when one follows up on references in publications where this statement is made, it turns out to be impossible to find this overwhelming evidence. More about this later. And there indeed are deniers within the medical profession, i.e. those who deny the validity both the causality and association aspects of the Cholesterol Hypothesis, and their voices are occasionally heard, mostly in one of several peer-reviewed British medical journals or in books one can buy at Amazon.com. The quotation given above represents the strong opinion of one high profile denier. This Review will attempt to critically evaluate the merits of the position taken by the deniers, in particular with regard to atherosclerosis.

The cholesterol hypothesis goes back a long way, in fact to the mid-19th century in Berlin where Rudolf Von Virchow found plaques in arteries of cadavers and observed that they contained cholesterol—an amazing observation given the development of medical science at the time. He made no connection with heart disease or heart attacks since these problems did not exist or were not recognized then. In fact the first medical description of a myocardial infarct (heart attack) as a clinical pathologic entity did not appear in the medical literature until 1915. Fifty years after Virchow's observation for some reason or other a Russian named Anitschkov fed rabbits a diet high in cholesterol and observed that their arteries thickened and filled up with cholesterol. But rabbits are not carnivores and cholesterol is totally foreign to their natural diet. Rabbits do not normally eat meat, eggs or milk products. It is doubtful that such studies are at all relevant to humans. After the WWII more cases of heart disease were being identified and there was growing interest as to both the cause and possible therapies. The breakthrough for the Cholesterol Hypothesis came with the study of Ancel Keys from the University of Minnesota, the instigator of the famous "Seven Countries" study of the relationship between heart disease, serum cholesterol levels and fat intake. As was eventually pointed out, Professor Keyes actually selected seven countries from a larger set in order to support his hypothesis. As Dr. Malcolm Kendrick, M.D., in his book The Great Cholesterol Con points out, by selecting a different set of seven countries using data available to Keys, one can get exactly the opposite correlation. Also, if one uses all the data the correlation disappears. Keys' work was followed by the famous Framingham Study which provided evidence, albeit rather weak, for a connection between cholesterol and coronary heart disease (CHD), particularly in young men, and by additional studies where rabbits were fed a diet containing cholesterol and fat, and the Hypothesis was well on its way to becoming enshrined. The final pillar was added when it was shown that lowering cholesterol in patients who had suffered a heart attack reduced the risk of a second heart attack (secondary prevention). Add to this the high rate of coronary heart disease among those with a familial predisposition to very high cholesterol levels even at a young age, and the story was complete. Circulating cholesterol caused heart disease and since heart disease involved atherosclerosis, cholesterol must be involved in the development and progression of atherosclerosis as well. Case closed. The evidence was compelling. No reasonable person could conclude otherwise. The end result has been a 55 billion dollar (U.S.) worldwide business in cholesterol lowering drugs, numerous successful careers in academic medicine, and a Nobel Prize for two researchers. Perhaps more importantly, there arose a widespread fear among many individuals of high blood levels of cholesterol as well as dietary cholesterol and fat and the result was psychological stress and in some cases an adherence to diets that may in fact be unhealthy. In addition, a hostile environment now exists for anyone who even suggests that there may be serious problems with what are almost universally viewed as the evidence-based foundations of the Cholesterol Hypothesis.

Today LDL cholesterol or its oxidized form largely replaces total cholesterol (TC) as the agent viewed as responsible for causing coronary heart disease (CHD). But TC and LDL go hand in hand. LDL levels are generally not directly measured but calculated from measurements of triglycerides (TG), HDL and TC. The calculation assumes that the TC can be regarded as the following:

• TC = LDL + HDL + 0.2 X TG (values in mg/dL)
• TC = LDL + HDL + 0.46 X TG (values in mmol/L)

In fact the statin cholesterol lowering drugs impact a number of biological processes and this has become a hot research area. Also a number of problems exist which are related in part to inexplicable dose and level dependencies. For example, consider studies that have different initial LDL levels and produce similar percentage LDL lowering. These studies all give similar risk reduction, but the endpoint LDL in one study can be higher than the initial level in another study, and yet the same benefit accrues. There have been a number of such studies and taken together they suggest that the initial and terminal levels of LDL have nothing to do with the risk reduction. A review in 2007 put the matter this way when discussing the use of statin drugs to lower cholesterol: "The relative risk reduction is approximately 20-40% regardless of age, sex, pretreatment level of LDL-C, race or preexisting myocardial infarction" [1]. Such results have caused a number of researchers in this field to look at the possibility that the statin drugs prevent recurrent heart attacks by some other mechanism than cholesterol lowering, and the levels of either TC or LDL are in a large part unrelated to the benefit of the drug treatment. This is now a very active area of research and that alone should suggest just how unacceptable is the argument that the Cholesterol Hypothesis is proven in terms of a causal relationship by the fact that benefits, mostly in secondary CHD outcomes, are produced by cholesterol lowering by drugs. Proposed non-lipid lowering mechanisms include improvement of endothelial dysfunction, reduced inflammatory response, stabilization of atherosclerotic plaques and reduced thrombogenic (clot formation) response [1,2]. It turns out that statins, which block a critical step in the biosynthesis of cholesterol, also eliminate two precursors to a number of biologically important molecules which in turn may be related to the above non-lipid lowering effects of these drugs. Lipid lowering will be the subject of the third review in this series where more documentation will be provided. But the essential point is that there are problems, both logical and factual, with the standard argument that because there is a decrease in risk of CHD when statins are used to lower TC and LDL, that it therefore follows that TC and in particular LDL cause CHD, or atherosclerosis for that matter. It is also of interest that there does not appear to be an explanation for just how LDL is supposed to cause atherosclerosis and coronary heart disease [3]. Rather, it is "work in progress." In what follows we will look at a number of studies that are either inconsistent with the Cholesterol Hypothesis or appear to actually falsify it.

AUTOPSY STUDIES
If circulating cholesterol causes atherosclerosis and thus coronary heart disease, one might expect to see a correlation between the extent of atherosclerosis and cholesterol levels. The first significant study appears to have been reported back in 1936. Two pathologists from New York University, K. Landé and W. Sperry, studied a large group of individuals who had died from violent incidents [4]. They examined the extent of coronary atherosclerosis observed at autopsy and found no correlation with serum cholesterol levels. Some dismissed these results by claiming that cholesterol values measured after death were not a reliable measure of levels while alive. But other studies enable one to discount this objection. A Canadian study examined a large number of veterans at death [5]. Adequate pre-death cholesterol data were available and levels varied considerably among the individuals but for any given person, they were fairly constant. Autopsy studies on all the veterans who died revealed no connection between the degree of atherosclerosis and blood cholesterol levels. The same results were found in a study from India. Mathur and coworkers [6] studied the changes in cholesterol levels subsequent to death and found them to be stable for at least 16 hours. Thus samples collected shortly after death, as was done by Sperry and Landé, were representative of pre-death levels. Next, Mathur's group studied 200 individuals who had died in accidents but were free of any preceding disease. No connection was found between cholesterol values and the degree of atherosclerosis. These studies involved what amounted to random selection. In other studies that also were random, similar results were reported [7].

The Framingham investigators also looked at this question. They found a very weak correlation between cholesterol levels and atherosclerosis at autopsy. The correlation coefficient was 0.36. Correlation coefficients of this magnitude generally accompany scatter plots where one can barely detect anything other than a random array of points. In fact, those trained in the physical sciences are generally appalled by the significance attached to small correlation coefficients in other branches of science. Also, in the Framingham cohort at that time, there were 914 deceased individuals, but the Framingham investigators selected only 127 (14%) for the purpose of studying atherosclerosis and cholesterol. Thus apparently this was not a random selection and the report did not describe the selection criteria. Did only 14% of the families involved allow an autopsy? Two studies from Japan claimed a positive correlation, but correlation coefficients were even smaller than found in the Framingham study, and in one study, the correlation appeared only in individuals with low or normal cholesterol levels, and in the other only in the elderly. Also, for those with very high cholesterol, the degree of atherosclerosis was the same whether they were young or old. In a study from Norway, claimed to support the Cholesterol Hypothesis, many people with normal coronary arteries had cholesterol levels as high as those for whom all three coronary vessels were constricted, and those with two constricted vessels had lower levels than those with just one constricted artery [8].

Thus the autopsy studies either do not support at all the connection between circulating cholesterol and the degree of atherosclerosis, or they produce such inconsistent results or very weak correlations as to cast serious doubt on the validity of the hypothesis. And after all, these studies go rather directly to the heart of the matter (no pun intended) by looking at actual atherosclerosis in dissected coronary and other arteries.

CALCIUM SCORES
The use of electron-beam tomography of coronary artery calcium (EBT CAC screening) has become a popular method for determining the extent of plaque formation and thus the degree of coronary atherosclerosis. There is even a popular book for the layman with the catchy title Track Your Plaque which has no doubt motivated many people to go out and get a so- called calcium scan. The results of the scan are generally expressed as a calcium score (CAC score), invented by Arthur Agatston M.D., a cardiologist better know to the general public as the author of the best selling book The South Beach Diet. As might be expected, there have been studies directly or indirectly addressing the simple question, is there a correlation between the calcium score and cholesterol levels. After all, if high cholesterol levels cause atherosclerosis (and strongly motivate treatment to lower them), then one might expect to see higher calcium scores associated with high levels of circulating cholesterol. The following studies address this issue:

• In a study reported in 2003, 5635 men and women aged 30-76 had CAC score determinations and were followed for an average of 37 months. The positive association between the adverse CHD event frequency and CAC score was not modified by the presence or absence of elevated cholesterol, suggesting no correlation between cholesterol levels and the CAC score and thus the degree of atherosclerosis [9].
• In a study of 6086 men and women of mean age 56-58, for men the CAC score was independent of LDL or TC. For women, only a very minimal CAC score was observed for LDL > 160 mg/dL and TC greater or equal to 240. For both of these categories, the mean calcium score was 1.0, i.e. negligible [10]. CAC scores in general range from 0 to over 400.
• In the Rotterdam Coronary Calcification Study, which involved 2013 men and women age greater or equal to 55, after exclusion of subjects on lipid lowering drugs, no association between TC and CAC score was found for men but one was found for women [11]. Nevertheless, women had a mean CAC score that was 1/6 those of men and was quite low.
• A study reported in 2005 compared Japanese and American men, aged 40-49 by determining the CAC score and parameters which included TC and LDL. While TC and LDL were higher in the Japanese cohort, only 13% of the Japanese men but 47% of American men had CAC scores > 0. In addition, when men from the two countries with CAC score > 0 were compared, there were no significant differences in either TC or LDL [12].
• In a study of 546 Brazilian men, when those with CAC score greater or equal to 75th percentile were compared with those in the < 75th percentile, only slight differences in TC and LDL were found [13].
• The association between racial differences, lipoprotein and lipoprotein particle size, and CAC score was examined as part of the HEART SCORE study. To quote the authors, "we found no significant association between lipoprotein or lipoprotein particle size and the extent of sub- clinical atherosclerosis as measured by CAC, whether in whites or blacks" [14].
• A large study determined CAC scores for over 22,000 men and 8000 women. Hypercholesterolemia was defined as TC > 200 mg/dL or the use of lipid lowering agents. The CAC scores were stratified into the ranges of 0, 1.0-9.9, 10-99.9, 100-399.9, and greater or equal to 400. For men the percentage with hypercholesterolemia was essentially constant (40- 42%) for the three upper CAC ranges and for women the percentages for these three score groups were 49-52%. Thus there was essentially no association between elevated TC and CAC scores over a huge range from 10 to > 400. Unfortunately, these results were not stratified by statin use [15].
• In a study published in 2005, over 4900 asymptomatic persons aged 50-70 were scanned for coronary calcium. It was found that CAC scores predicted CAD events independent of the standard risk factors and in fact more accurately than the standard risk factors or C-reactive protein levels. This study found no correlation between CAC scores and LDL levels. In this cohort, LDL levels were 143 plus or minus 33 (standard deviation, not range) and TC was 224 plus or minus 33 mg/dL. Thus the range of levels, which is greater than the standard deviations, were large enough to provide a meaningful test of the association between cholesterol levels and CAC scores [16].
• Hecht et al [17] examined the correlation between serum lipids and CAC scores in over 1000 consecutive asymptomatic individuals referred for EBT. They found that TC, LDL, HDL and the TC/HDL ratio did not correlate with either the prematurity or extent of calcified plaque burden.
• In a large multi-ethnic study, Kronmal et al [18] found only a very weak to insignificant associations between LDL and HDL and the change in CAC scores over time, i.e. the progression of atherosclerosis.

Thus 10 studies mostly fail to find a significant or clinically meaningful correlation between an established measure of the extent or progression of atherosclerosis and circulating TC or in some cases LDL. In fact, other risk factors such as smoking and hypertension did indeed correlate in many studies, but cholesterol never made the grade. Thus the electron-beam tomography studies are consistent with the autopsy studies, which is gratifying since, if we ignore small differences between visually and EBT identified plaques, both approaches are looking at more or less similar pathology.

CORONARY ANGIOGRAPHY
Coronary angiography involves inserting a catheter into the femoral artery in the groin and pushing it up through the aorta until it reaches the coronary vessels. A contrast medium is then injected to allow imaging of the individual coronary arteries and these images can reveal blockage attributed to atherosclerosis. Thus studies can examine the correlation between serum cholesterol and angiographically identified deposits and narrowing in the coronary arteries. But coronary angiography is not without its morbidity and mortality and is generally performed only on individuals with at very high risk or with severe symptoms of heart disease who are young or middle-aged. Thus these studies are not representative of the asymptomatic public. In addition there will be some patients who have familial hypercholesterolemia. This would introduce a bias since, as will be discussed below, it is not at all clear that the atherosclerosis that accompanies this syndrome can be compared to that found in individuals who do not have the mutation. Any study heavily weighted with individuals having very high cholesterol levels will be confounded by the potential presence of subjects with this syndrome. Also, in many studies, cholesterol lowering drugs were being used, further confusing the question.

Thus if the issue is the Cholesterol Hypothesis, angiographic results do not appear to be a very good way to study its validity. On the other hand, both the autopsy and calcium score approaches allow the examination of asymptomatic individuals which is much more to the point and less subject to confounding and bias. Nevertheless, as Ravnskov [8] discusses at length, angiographic studies that examined the relationship between cholesterol levels, their changes, and the presence and progression of atherosclerosis frequently found inconsistent results where progression occurred in the presence of both increasing and decreasing cholesterol levels. In his words, "to prove that high cholesterol is the villain—not just an innocent bystander—demands that a change in the cholesterol concentration for each individual is followed by a change in the degree of atherosclerosis in the same direction. But in all studies these changes occurred haphazardly." Also, most of the results are presented as the constants of correlation equations and almost always the degree of correlation is very poor (low correlation coefficient).

HOW ABOUT FAMILIAL HYPERCHOLESTEROLEMIA (FH)?
Individuals with FH have very high TC and LDL due, it is thought, to what is called an LDL- receptor deficiency which results from a mutation. An argument for the Cholesterol Hypothesis involves claiming that members of such families run a great risk of dying form CHD at an early age. Ergo, elevated TC and LDL cause atherosclerosis and CHD. To use individuals with this mutation as "proof" of the Cholesterol Hypothesis requires that the only difference between the FH people and the rest of us is that they produce huge excesses of TC and thus LDL. But it is not that simple. Individuals who inherit from both parents not only have highly abnormal levels of cholesterol in their atherosclerotic deposits, but also in other organs. Cholesterol levels can go to 1000 mg/dL or higher. And lowering their cholesterol levels drastically does not reverse their atherosclerosis. Also, these individuals have blood-clotting abnormalities which may responsible for the elevated rate of heart attacks. It is also argued by some that the nature of the atherosclerosis is different in FH as compared to non-FH individuals [19]. This does not seem to be an ideal group to use in justifying a hypothesis regarding the cause of atherosclerosis or CHD. Nevertheless it is one of the major pillars upon which the hypothesis rests.

An interesting and perhaps unique study from the Netherlands relates to this question. A large pedigree was traced back to a single pair of ancestors in the 19th century and a family tree mortality study conducted which started in the early 1800s. All members had a 50-50 chance of carrying the mutation for familial hypercholesterolemia. Mortality data over the full time span up to modern times was available for this large group as well as for the corresponding general population. Overall mortality was not increased in carriers of the mutation during the 19th and early 20th century. The mortality then rose reaching a maximum between 1935 and 1964. The authors comment that the risk of death varied significantly among patients with FH and that this was, in their opinion an indication of strong interaction with environmental factors. One can of course ask why for over more than a century no excess mortality showed up because of the elevated TC and LDL levels if in fact these lipoproteins cause atherosclerosis and CHD.

Kendrick [3] suggests that perhaps lipoprotein (a) is involved since there seems little doubt that this protein is elevated in those with the FH mutation and that this protein is regarded as a strong independent risk factor for CHD. However, he finds the strongest argument against FH causing CHD is that most people who die with heart disease do not have highly elevated LDL levels and most who have these LDL levels do not die of heart disease, even people with FH.

THE MONICA STUDY
This is an acronym for a huge World Health Organization study of cardiovascular disease. Among other things, the association of CHD deaths and TC was examined for a large number of countries. In one looks at a plot that displays the results from this set of countries, two things jump out at you [8]. First the data, which clusters between TC of about 210 and 250 mg/dL and covers a CHD death rate from some very low number to over 450 events per 100,000, shows that no matter what the cholesterol level is in this range, both very high and very low rates of CHD mortality are found. At a level of about 225 mg/dL the mortality for a number of countries ranges from 70 to 427 deaths per 100,000. Also, when data from several sites within a country are provided, there is a large variation in mortality at a given TC level. Overall, there is no apparent correlation between CHD deaths and TC, contrary to what would be expected on the basis of the Cholesterol Hypothesis. Just two countries are outside the main scatter, China and Japan, and both have both low TC and low rates.

Japan merits a bit of discussion. The Japanese living in Japan in general had both low cholesterol levels and low rates CHD mortality, but immigrants to the U.S. had high cholesterol levels and had CHD mortality comparable to Americans. Convincing proof of the Cholesterol Hypothesis. But let's dig deeper and look at what a British physician found during his Ph.D. research. He looked at the relationship between TC levels and social factors, eating habits and lifestyle among the immigrant Japanese. He found conclusive evidence that it was not the food that raised the cholesterol of the Japanese immigrants, or that elevated cholesterol values increased their risk of CHD death. Rather, he found that those who maintained their cultural traditions were protected against heart attacks, even though their cholesterol increased as much as in the immigrants who adopted the Western lifestyle and diet and who died from CHD at a rate comparable to the Americans. In fact, the Japanese who preferred lean Japanese food but adopted other aspects of the American way of life had CHD twice as often as those who maintained Japanese traditions but preferred standard American diet [20,21]. These studies, which contradict the Cholesterol Hypothesis, have been largely ignored, in spite of having been published in a high profile peer-reviewed American journal. Thus, if the low rate of CHD among the Japanese has little to do with cholesterol levels and the low mortality is for other reasons, then the MONICA result for Japan are explained. This leaves only one discordant point in an otherwise apparently random scatter of points in the Monica results.

The relationship between cholesterol levels and mortality is actually both complex and fascinating and will the subject of Part II of this review.

FRAMINGHAM RISK SCORES
Framingham is a small town near Boston, MA, and has been the site of a study involving a large number of its inhabitants. The study has been ongoing since 1950 and now even involves the children of the original cohort. Early results from this study had a major impact on the development of the Cholesterol Hypothesis. A commonly used graphical representation of the results regarding cholesterol and the incidence of CHD shows two curves where the percentage of participants with or without CHD is plotted against TC. Both curves rise from around 100-120 mg/dL TC, pass through maxima percentage at around 200-220 mg/dL, and decline to zero at 400 mg/dL except for a small bump at higher TC attributed to those with FH. The CHD patient's (n = 193) curve is slightly displaced to higher TC compared to that of people without CHD (n = 1378). While believers in the Hypothesis point to the added risk associated with elevated TC reflected in the displaced curve, its can also be pointed out that the vast majority of patients with CHD had TC levels similar to participants without CHD, and that the added risk appeared marginal. What this often displayed plot fails to show is that the risk of overall mortality associated with TC disappeared for men above age 48. In addition, in a longer follow-up, for participants whose cholesterol had decreased on its own (i.e. no lipid lowering treatment), for each 1% drop of TC there was an 11% increase in coronary and total mortality [22].

The Framingham study also gave rise to the so-called Framingham Risk Score, a risk estimate of having CHD during the next 10 years. This score is widely used in the office setting to assess an individual's risk of CHD. If one looks at the way the score is calculated, as one gets older, age become by far the dominant factor with the importance of cholesterol dropping off dramatically with age until it becomes almost insignificant for men and makes only a slight contribution for women. This is in spite of the arterial exposure to TC and LDL obviously increases with age. This is not a picture of overwhelming support for the Hypothesis.

WHAT ARE THE REAL RISK FACTORS FOR ATHEROSCLEROSIS?
If we use CAC as a surrogate for atherosclerosis, then given that there does not seem to be any connection with serum cholesterol or LDL, are there other traditional or non-traditional risk factors that correlate with plaque burden? Traditional risk factors that consistently turn up as most important in studies of the extent of atherosclerosis are smoking, hypertension, gender and age [13,17,23,24]. However, when the correlation between the Framingham 10-year risk score and the CAC score is examined, generally a very poor correlation is found. In one study, 20% of individuals with very low Framingham risk (less than or equal to 9% 10-year risk) were in fact found to have advanced atherosclerosis as judged by their calcium scores [25]. This study also found that the ability of the CAC score to predict advanced atherosclerosis was improved by adding family history of heart disease, obesity and physical inactivity to the traditional risk factors. But it appears likely that there is still a major factor is being omitted. It is possible that this factor is chronic stress and depression. There is also growing evidence that chronic stress and depression correlate with the extent of coronary calcified plaque [26-28]. Furthermore there is very good evidence that stress in general is a strong predictor of CHD and CHD events [29,30]. In one very large study an attempt was made to identify the major potentially modifiable risk factors for a heart attack. It was found that the major factors were a particular blood lipoprotein (apolipoprotein), smoking, hypertension, diabetes and psychological factors, all of which had approximately equivalent importance [30]. Having psychological factors ranked approximately equal to diabetes, both of which increased the risk of a heart attack by a factor of 2 to 3, suggests that chronic stress and depression may be major factors not seriously considered in routine CHD risk assessment, especially in the office setting. Its omission may account for a part of the failure of sets of risk factors such as the Framingham score to correlate with the extent of atherosclerosis as determined by CAC.

CONCLUSIONS
What we are seeing is that over the years there has been a steady flow of problems and "how come" questions that eat away at the credibility of the cholesterol hypothesis. The resolution of all of these problems is simple—most features of the hypothesis are false. Rephrasing the hypothesis such that it only states that atherosclerosis and CHD are associated with cholesterol levels removes what is considered an incorrect attempt to connect observational studies with causality, but the evidence presented above also argues equally well against there being, for the most part, any such association.

Part II - CHOLESTEROL AND MORTALITY

Mainstream medicine regards the cholesterol-heart disease connection as an established fact. It has been elevated almost to a sacred belief. Anyone who questions the foundations of this hypothesis runs the risk of being branded a heretic or someone who is unable to appreciate the wisdom and beauty of one of medical science's outstanding achievements. Many careers have been built on the Cholesterol Hypothesis, careers replete with high profile medical academic positions, drug company supported lectureships and consulting and financial support for research. Any medical professional who openly expresses serious doubts runs the risk of being ostracized by his or her peers. But there are indeed deniers and sceptics, and it seems important that Newsletter readers are aware of their views [1,2]. Avid readers of the British journals The Lancet, the Quarterly Journal of Medicine and the British Journal of Medicine are probably among those most keenly aware of these voices crying in the wilderness. Such criticism or questioning is rare in North American Journals. But science thrives and progresses on dissent, controversy and attempts to falsify hypotheses, a fact that seems underappreciated by those who guard the conventional wisdom. Some would argue that this does a profound disservice to the progress of medical science and the ultimate discovery of so-called truth.

Painting cholesterol and in particular LDL as demons, sort of on a par with toxic substances or even pathogenic bacteria, is unfortunate since this almost totally obscures the fact that cholesterol is essential for our wellbeing. It is involved in ensuring the integrity of cell walls and in the synthesis of testosterone, estrogen, dehydroepiandrosterone (DHEA), progesterone, cortisol, and last but not least, vitamin D by photosynthesis from the exposure to skin to ultraviolet light. We produce cholesterol in the liver, and in general, high dietary consumption results in lower endogenous production, and many feeding studies have found virtually no dependence of serum levels on dietary intake [3]. This is obviously inconsistent with the conventional wisdom which suggests limiting the dietary intake of cholesterol, a recommendation which the food industry picked up on and has played to the limit. Finally, the Cholesterol Hypothesis has resulted in widespread screening and a vast amount of anxiety, stress and pharmaceutical intervention over cholesterol levels labeled elevated and declared dangerous and even life-threatening.

What we will call the Cholesterol Hypothesis simply states that that high levels of total cholesterol, i.e. TC and LDL cholesterol (LDL), cause atherosclerosis and are associated with elevated risk of developing coronary heart disease. The word cause is important in this context because cause is relatively difficult to establish in many human disorders and CHD appears to be one of them. The reason in part is the extraordinary complexity of the sequence of events associated both with the development of atherosclerosis and the events leading to an acute coronary episode, e.g. a heart attack. One thing that appears clear is that the simplistic view of cholesterol clogging up ones arteries (the kitchen drain analogy) is just that—simplistic. The evidence against the Cholesterol Hypotheses as regards cholesterol-causing atherosclerosis was reviewed in Part I.

STUDIES CONCERNING CHOLESTEROL AND MORTALITY
Part II of this review series will deal with the question of cholesterol levels and both overall mortality and coronary heart disease mortality in men and women of all ages who are free of coronary heart disease as evidenced by the absence of chronic or acute angina, a history of a heart attack intervention to open an artery or insert a stent or coronary artery bypass. One frequently sees the concern put forward in the medical literature that some intervention or procedure has not been proven to favorably influence mortality and thus its value is questionable until this aspect is settled in randomized trials. In fact, critics of screening frequently use this as a "gold standard." Some consider the impact on mortality to be an important if not essential factor in the risk-benefit equation and in addition, there is always the possibility that an intervention or procedure actually increases mortality. In addition, impact on mortality is viewed by some as important in judging risk factors. Overall mortality is a relatively easily established endpoint because there is little room for debate since the patient is dead, but disease specific mortality is another matter since in may not always be clear as to what was actually the cause of death and death certificates can be inaccurate or even simply wrong about this.

One of the largest and most recent studies to address the mortality issue was published in 2004 and involved almost 150,000 Austrian men and women ages 20 to 95 years [4]. The study involved multiple evaluations of total cholesterol over a 15-year period between 1985 and 1999. Overall (all-cause) mortality and CHD mortality were evaluated by comparing the lowest and highest quartiles of TC with the middle quartiles used as reference. The following results are of particular interest.

• For men, there was no statistically significant association between all-cause mortality and high TC ( > 248 mg/dL) for age > 50 years. For high TC, there was a weak association for ages < 49 years. For low TC (<187 mg/dL) there was an increase in the mortality rate.
• For women, there was no statistically significant association between high TC (> 244 mg/dL) and all-cause mortality at any age. For low TC (< 184 mg/dL), there was an increased risk of all- cause mortality for ages >49 years of age.
• For men, high TC was significantly associated with CHD mortality for the age group 20-49 years, and a weak positive association was also found for those greater or equal to 65 years.
• For women, high TC was associated with weakly elevated CHD risk only in the age group 20- 49. However, some studies discussed below failed to find this enhanced risk.
• For both men and women, there was no association between either high or low TC and stroke mortality, but low TC was associated with increased risk of cancer mortality in men 50-64 years and women of >50 years of age.

Thus for overall mortality, high TC was not a significant issue for men over 50 or for women at any age, but in fact there was evidence of increased overall mortality associated with low cholesterol. For men under 50 years of age, overall mortality exhibited a U-shaped curve vs. TC. For women, the mortality rate curve which has been seen in many studies is more or less flat at high to intermediate TC levels and then slopes upward indicating increased mortality associated with low TC.

The weak positive association between CHD mortality and high TC for men over 65 is inconsistent with a large number of earlier studies. Ravnskov has summarized these studies in a recent commentary [5]. His review summarized 13 studies, the largest 4 of which involved over 12,000 men and women, where 7 studies found that the lower the TC or LDL, the higher the mortality, and all 13 studies found that high TC or LDL did not predict increased mortality in this age group. Seven of these studies had subjects in the age group 60-65 years and one exhibited an inverse relationship where increasing TC decreased coronary mortality (this was actually part of the famous Framingham Study).

Similar results related to elderly men were reported for the Honolulu Heart Program follow-up study which involved over 3500 men followed for a maximum of 20 years [6]. The age range was 71-93 years. Mean cholesterol levels for the quartiles were 149, 178, 199, and 231 mg/dL. When the lowest quartile for TC was used as reference, the relative risk for all-cause mortality decreased (0.72, 0.60 and 0.65) as TC levels increased. Thus there was benefit rather than risk associated with high TC as it relates to all-cause mortality. When results from the first year of follow-up were excluded, similar risk reduction with increasing TC were found. This argues against the low levels being due to preexisting illness. In addition, the long-term follow-up in this study, in the opinion of the authors, renders the hypothesis untenable that the low TC effect is due to undiagnosed preexisting conditions that reduced TC levels. The authors state that, "We have been unable to explain our results." This could be translated into a statement that the results are not in accord with the conventional wisdom.

In a U.S. study involving over 10,000 men and 8600 women, the relationship between TC and all- cause mortality was determined over a follow-up period of 22 years [7]. There was no stratification by age. For women, there was no significant association with overall mortality and TC. A third of the women had TC < 201 mg/dL and a third had TC > 240 mg/dL. For men, the only positive association was for TC > 240 mg/mL, but the relative risk was the lowest among all the risk factors that gave a positive association (e.g. blood pressure and smoking).

In a large study of Korean men aged 30-65, it was found in a 6.4 year follow-up a low cholesterol level (< 165 mg/dL) was associated with increased risk of all-cause mortality [8]. The risk of CHD mortality was found only for men with the highest cholesterol levels (greater than equal to 252) but there was no stratification by age. Thus this study also found a U-shaped relationship between overall mortality and TC. The strongest disease specific increase in mortality with decreasing TC levels was seen for cancer. The authors rule out the possibility that this was due to preexisting cancer since attempts were made to exclude patients with cancer or precancerous condition at baseline, and the increased risk of cancer with decreasing cholesterol levels remained significant throughout the 5-year follow-up period. This is an important point since the conventional explanation offered by mainstream medicine for the increased cancer mortality with decreasing cholesterol levels is that it is due to a preexisting condition that resulted in the low level. Also, the absence of increased cancer incidence in studies where cholesterol was aggressively lowered with statins in studies for secondary prevention or with very high risk individuals does not appear relevant to the above issue since the patient population was different and there is the possibility that the statin drugs have anti-cancer properties unrelated to cholesterol lowering.

Studies published very recently present the same picture. Mortality was significantly correlated only with a low level of TC (less than equal to 160 mg/dL) in a small group of men and women 84 years or older [9]. In as study of over 300,000 Korean women age 40-64, no significant association was found with TC and CHD mortality for the 40-55 age group of pre- or postmenopausal women. For the 56-64 age group, the only significant association was for TC levels greater than equal to 236 mg/dL, i.e. the highest quartile. Given that all other associations were non-significant, this single result may be a fluctuation [10]. Finally, in a study of individuals presenting with ischemic stroke (from a blood clot) it was found that higher cholesterol levels favored minor strokes and thus post stroke mortality was inversely related to cholesterol [11]

Stamler et al examined the question of the relationship between TC and all-cause mortality and CHD mortality by an analysis of three prospective studies involving men 39 years or less of age [12]. One study had a mean age of about 30, another 32 the third study 37 years. In men in this age group, the overall and CHD mortality risk was elevated when those with elevated TC levels were compared to those with TC < 160 mg/dL but in the largest of the three studies which involved over 60,000 participants, significant increased in all-cause mortality was seen only for levels above 230 mg/dL when TC was stratified into quintiles. The two other studies revealed CHD mortality risks in all quintiles above the reference quintile (< 160 mg/dL). These results are consistent with studies discussed above where the data was stratified by age. The authors conclude that these results suggest a longer life expectancy for younger men with favorable cholesterol levels.

Thus there appears to be little data to indicate that elevated cholesterol levels are positively associated with CHD or overall mortality and in fact the opposite appears fairly well established, low cholesterol levels are accompanied by elevated mortality. Young men appear to be an exception.

CHOLESTEROL AND SUDDEN CARDIAC DEATH
In the U.S. population, about 65% of the cardiac related deaths in 1998 in adults aged > 35 were due to so-called sudden cardiac death (SCD) [13]. As de Lorgeril and Salen point out in a recent short paper in the journal Nutrition, Metabolism and Cardiovascular Diseases [13], two studies indicate that high cholesterol levels are not a risk factor for SCD. They point out that these findings are surprising given the widespread view that high cholesterol is a major risk factor for CHD death. In one study, over 121,000 women aged 30-55 with high cholesterol were followed [14]. The other study involved men with a mean age of about 60 (range 40-84) and a range of TC of 196-247 mg/dL [15].

Thus the picture emerges that if cholesterol levels are considered in the context of overall or CHD mortality high cholesterol is only an issue for young men and appears protective for women of all ages and men over 50 years of age. However, this applies to individuals who do not have coronary heart disease.

The current guidelines [16] of the National Cholesterol Education Program for TC (mg/dL) are:

• Desirable - < 200
• Borderline High - 200-239
• High - greater than equal to 240

It seems noteworthy that these guidelines are not stratified by either age or gender. While the guidelines emphasize LDL, to a large extent TC is a surrogate for LDL in that high TC almost always implies high LDL. The guidelines also focus on risk of CHD rather than mortality, but the issue being discussed here is in fact mortality and whether or not serum cholesterol levels are related to either CHD mortality or all-cause mortality. However, the guidelines do introduce age and gender, but only when the estimation of risk is with the Framingham risk factor calculator. In one version of the Framingham data, risk for CHD includes fatal or non-fatal heart attack and unstable angina. In another version fatal and non-fatal heart attacks are the only endpoints.

HOW ABOUT THE ENHANCED RISK OF CHD IN YOUNG MEN WITH ELEVATED TC?
It seems obvious the there is an anomaly associated with young men and the risk of CHD mortality associated with elevated TC. Why for example, does the risk not persist in older individuals where the exposure to circulating cholesterol is much longer? It does indeed appear to be true—high cholesterol predicts CHD in young and middle-aged men. But if cholesterol is merely a marker and plays no role in causative mechanisms, then the problem posed by young men can be resolved if some mechanisms unique to this age and gender group are responsible for cholesterol elevation. It has been suggested [5] that one factor may be stress which is well known to elevate serum cholesterol levels, and which would be expected to peak in age period prior to 50. Ravnskov [5] suggests a number of factors which can be expanded upon. This is the age where most men are in the midst of their professional careers, in many cases subject to great stress, even harassment. After all, it is well known that many individuals hate their jobs, their boss, the part of the country where they work, etc. There is the ever present risk of being fired, making mistakes that lead to bankruptcy, loosing ones job because of downsizing or failing to achieve or get promoted or get that much desired executive position, etc., etc., etc. There is the worry of loosing ones job at an age where it would prove difficult if not impossible to get an equivalent one. This is the age where marital break-ups are common and are generally highly stressful, where teenage children drive their parents mad or frantic with worry. It is when ones children may succumb to drugs and alcohol, and the age where parents sometimes have to contend with a child's unwanted pregnancy. One could go on indefinitely with the potential horrors that might confront the middle age man.

There have been a number of studies that examined the association between cholesterol levels and work-related stress. While it is true that the higher the stress the higher the serum levels, the magnitude of elevation is small and does not seem sufficient to support the idea that this is the mechanism giving rise to the anomalous effects seen in young men, which incidentally includes a weak but significant correlation between CHD in general and TC. Thus there is a problem with Ravnskov's suggestion. However, studies of cholesterol and CHD events or mortality that correct for confounding have of necessity failed to take into account at all one confounder, the magnitude of blood pressure changes due to stress. If resting blood pressure is measured and used as a variable in adjusting the association between CAD and cholesterol, this ignores the fact that higher levels of cholesterol appear to be associated with enhanced or exaggerated response to stress (hyper-response) and hyper-responders have enhanced CHD risk in response to stress. Thus in the population of young men, there will be these hyper-responders who will on average have considerably higher cholesterol levels, but it is the hyper-response of blood pressure to stress that increases the risk and thus the action of cholesterol is indirect. The impact of large fluctuations in blood pressure occurring on a daily basis in response to stress, either work related or domestic, would be the direct but unmeasured association with increased risk of CHD mortality or events. This could account for the modest correlations between CHD mortality or CHD events and cholesterol in young men who are exposed to a higher level of stress than older men and who may cope with stress less successfully than women. If studies had adjusted for the magnitude of blood pressure response to stress challenges it is suggested that the modest connection between cholesterol and CHD risk might disappear [17]. But operationally this appears impossible in studies large enough to provide adequate statistics since all that is practical to measure is resting blood pressure and it is obviously impossible to duplicate in a clinical examination or screening session the day- to-day stress of a toxic workplace or the stress associated with domestic problems.

In his recent book The Myth of Cholesterol, Dr Paul Dugliss, M.D. makes a case for stress being so important in the etiology of CHD that it renders cholesterol to a position of near irrelevance. He compares typical estimates of risk reduction due to stress reduction with risk reduction from cholesterol lowering drugs. The former can result in risk reductions of up to 540% for heart attacks, whereas typical risk reduction with lipid lowering is about 25% (but only in the subgroups where it works at all).

This view is strongly supported by results from the INTERHEART study, a case-control study involving subjects from 52 countries that examined the relative importance of risk factors for heart attacks [18]. It was found that psychosocial factors including stress at work and at home, general and financial stress, stressful life events, and depression resulted in risk of heart attack that was greater than diabetes or smoking, both of which are major traditional risk factors. Diabetes alone automatically confers a 10-year risk of CHD or CHD events of greater than 20%, i.e. high risk.

HOW ABOUT HDL, THE SO-CALLED GOOD CHOLESTEROL?
A rather small fraction of TC is due to HDL cholesterol, the so-called good cholesterol and TC is not a good surrogate for HDL. LDL according to the conventional wisdom is the bad guy, but some point out that this may only apply to the oxidized form, something that appears to be rarely measured in routine physical exams. Thus the following question: are the protective properties of HDL borne out in the relationship between HDL levels and mortality? Several studies have addressed this issue.

A study from Finland [19] looked at CHD mortality and HDL. The variation of HDL levels with age over the total range from 25 to 64 was very small and thus the results without age stratification are of interest. For both men and women, the risk reduction for CHD mortality was about 10% per 4-mg/dL increase in HDL and was statistically significant. This appears to be clinically relevant since the difference between high and low HDL is about 20 mg/dL.

In a study of type 2 diabetics, cardiovascular mortality rates were independent of TC or LDL but significant protection was seen with HDL levels greater than about 53 mg/dL. The follow-up involved over 10,000 person-years and age did not influence the protective effect of HDL. However, adjustment for a large number of potential confounders reduced the benefit in all but those >70 years of age [20]. Thus there is nothing in these studies which contradicts the belief that high levels of HDL are protective.

CONCLUSIONS
Ravnskov points out that if high TC or LDL were an important cause of CHD or CVD, it should be a risk factor for both genders in all populations and in all age groups. But in many populations the association between TC and mortality is absent or even inverse, i.e. just the opposite, where increasing TC is associated with lower coronary and total mortality. In the elderly high TC is associated with longevity in most studies. The results with the elderly are especially significant because both the highest mortality and the greatest incidence of CVD are seen in the elderly. Ravnskov advances the hypothesis that the beneficial effects of high cholesterol on the immune system may explain why sometimes an inverse association is found between TC and mortality and as well inverse associations seen sometimes which contribute to the inconsistencies that characterize the angiographic studies which attempt to link TC or LDL and atherosclerosis [5].

It is also noteworthy that the vast majority of acute coronary events such as heart attacks occur after the age of 60, an age where the question of cholesterol and either CHD mortality or overall mortality becomes an issue only in that high TC appears either neutral or decreases rather than increases the risk of death. Thus, for a large fraction of the adult population, concern about high cholesterol amounts to worrying about something that appears to have no bearing on CHD mortality, and perhaps even increase life expectancy.

There are millions of individuals taking cholesterol-lowering drugs. The number worrying about their high cholesterol must be very large indeed. The total drug company income from statins is approximately $55 billion (not a misprint) per year. Today, recommendations are mainly based on LDL cholesterol, medical history and the presence of traditional risk factors, two of which (Framingham) depend on TC and HDL. But levels of LDL are closely tied to TC. Thus the studies that examined mortality as a function of TC are relevant. The evidence or lack thereof presented above would suggest that the recommendation of these drugs for women and the elderly who are free from symptomatic CHD and have not had a heart attack is not based on evidence involving risk of either CHD mortality or all-cause mortality, and some regard this as the most important consideration. If low TC indeed increases mortality, then this becomes an additional consideration, one that probably never comes up in consultations since mainstream medicine dismisses all the data regarding this problem with arguments given above. In addition, in younger men, high TC may lead to inappropriate therapy if the real problem has to do with high stress. Also, obsession with cholesterol levels may result in avoiding a number of potential actions associated with primary prevention. Since statin drugs have side effects, the above considerations assume added significance.

Thus we have now seen in the first two parts of this series that there does not appear to be a connection between CHD mortality or all cause mortality and circulating cholesterol. Nor are cholesterol levels associated with the extent of atherosclerosis. In connection with this second point, the British physician Malcolm Kendrick, a long-time student of the Cholesterol problem, poses three key questions in his book The Great Cholesterol Con:

• Why don't veins develop atherosclerosis?
• Why does atherosclerosis develop in discrete (separate) plaques?
• If high LDL level causes atherosclerosis, how can vast numbers of people with low LDL levels get the same disease?

When one thinks about it, it is remarkable that atherosclerosis does not develop in veins. They have the same cellular wall structure as arteries and they are exposed to identical levels of TC and LDL. If an artery is replaced by a vein, as in bypass surgery, the vein now acting as an artery can develop atherosclerosis as frequently seen in bypass restenosis, but replace a vein with an artery and the artery appears protected against atherosclerosis. Kendrick concludes that this has something to do either the position in the body or the function of arteries. But how can LDL be the culprit since it remains constant throughout the circulatory system?

For the discrete plaque patch puzzle, he likens this to sunbathing and getting burned only in patches in spite of uniform exposure, the analogy being with the uniform exposure to the hypothetical causative agent cholesterol. If plaques form in damaged areas and LDL does not itself damage the arterial walls, then something else causes heart disease.

Defenders of the Hypotheses respond to the third question by simply saying the low LDL levels that still result in atherosclerosis are not really low but are in fact high, and that almost everyone has the disease of hypercholesterolemia and clearly need aggressive therapy. Mankind has evolved, perhaps even over the past century, such that they now have a deficiency disease, where the deficiency is a prescription drug. Critics tend to view this explanation as total nonsense. This view will be examined and documented in Part III of this series.

A simple solution to the problems posed by these three questions is that the Cholesterol Hypothesis is false.

Part III - CHOLESTEROL AND STROKES: LIPID LOWERING

STROKE RISK AND CHOLESTEROL
Epidemiologic studies have not shown a clear association between cholesterol levels and stroke. However, the subject is complicated by the presence of two distinct types of stroke, i.e. ischemic and hemorrhagic. The former involve occlusive interruption of blood flow while the latter are due to bleeding. In a number of studies and some meta-analyses, these have been lumped together. The problem that arises involves what appears to be an increase in hemorrhagic stroke at low cholesterol levels which would compensate for a decrease in ischemic strokes, drive the results toward the null and mask a positive association between stroke and serum cholesterol. The large meta-analysis of prospective studies which comprised 450,000 subjects with 13,000 strokes [1], and which found no association between incidence and cholesterol levels, has been questioned on these grounds by a number of observers. Law et al [2] attempted to resolve this question with a meta-analysis of observational studies where the ischemic and hemorrhagic strokes were reported separately. At issue were LDL levels. Seven studies reporting on ischemic stroke qualified, but one involved only smokers and is hardly representative of the general population. Furthermore, the paper referenced for the data on this particular study was not relevant because it did not stratify by LDL and in addition, the results for total cholesterol failed to yield statistically significant associations, whereas the LDL results used in the meta-analysis were given as significant. References to the other studies included in the analysis lead to papers that also did not measure LDL and the authors give no indication as to the source of the LDL data on which they base their analysis. Six of the seven studies found no significant association between LDL and stroke. All seven when subjected to a meta-analysis produced a statistically significant 15% decrease in risk of ischemic stroke for each 38-mg/dL decrease in LDL. The major goal of meta- analyses is to tighten up the statistics by reducing the uncertainty in the result through the use of a larger sample of subjects. The individual studies are generally weighted and this weighting process is arbitrary. In fact, there is in general a problem with the meta-analysis of observational studies. As Willett points out, issues of validity in observational or prospective studies are determined by confounding and bias, and this is not altered by the enhanced statistical power of a pooled data study or meta-analysis [3]. And it is the enhanced statistical power that produces a significant result from a set of studies that individually showed no statistically significant results. Law et al also found a 19% increase in hemorrhagic stroke for each 38% decrease in LDL, but the same problems described above apply here as well. Thus this meta-analysis does not seem to resolve the issue in question.

Two studies were not considered by Law et al, in one case simply because the study came after their paper was prepared. One, the Eurostroke Study, which examined both ischemic and hemorrhagic strokes in the general population using a case-control model, found no significant association between cholesterol and either stroke type, fatal or non-fatal [4]. A very large prospective Korean study reported in 2006 [5]. Involved were over 787,000 individuals which provided a large number of strokes for analysis. Six ranges of total cholesterol were used, with < 130 mg/dL as a reference. When the results were adjusted for confounding, only cholesterol levels grater than 270 mg/dL were associated with increased risk of ischemic stroke, and only 0.9% of the subjects were in this category. For everyone else, there was no statistically significant association. For hemorrhagic stroke, total cholesterol levels above 160 mg/dL were protective when < 130 mg/dL was used as the reference, i.e. as cholesterol levels increased, the risk decreased. However, an interesting connection with alcohol consumption was made. By using a blood marker for alcohol consumption, it was found that when this marker was low, the association between low cholesterol and enhanced risk of hemorrhagic stroke disappeared. Thus in apparently healthy populations, the balance of evidence suggests that the incidence of stroke is unrelated to serum cholesterol levels but someone with a very high level may be at a modestly enhanced risk.

The absence of an association between cholesterol levels and the risk of ischemic stroke was termed a paradox by Matthias Enders in a review published in 2005 [6]. But he points to an additional paradox. Given that cholesterol does not appear associated with the incidence of ischemic stroke, it is then curious that treatment with statin drugs reduces the incidence of stroke, at least in patients with existing CHD or a history of stroke. Several studies have shown this. When data was combined from 9 trials including over 70,000 participants with established CHD or high risk of CHD, the relative risk reduction was 21% with an absolute risk reduction of 0.9% (number of individuals needed to treat to prevent one stroke—111) [7]. When intensive (higher dose) statin therapy was compared with therapy using usual doses, a pooled analysis of 4 trials yielded an additional 18% relative risk reduction but only a ½% absolute risk reduction [8]. In older studies statins were not found to reduce stroke risk in typical populations without known CHD, but in these primary prevention studies the age was rather low and thus so was the stroke incidence, and these trials lacked the statistical power to reliably detect a significant effect. For asymptomatic individuals not deemed high-risk, the benefits of statins for primary prevention is not clear. But there remains the problem of whether or not the primary action of the statin drug in this context is related to cholesterol lowering or some other anti-stroke action. Enders points out that the observed dose dependence relative to baseline cholesterol levels suggests that cholesterol lowering may not be the important factor, and this view is bolstered by studies of cholesterol lowering using non-statin drugs where there was no decrease in stroke incidence in spite of significant cholesterol lowering [6].

There is some evidence that low cholesterol levels induced by statins increase the risk of hemorrhagic strokes [9-12]. In the just published SPARCL trial it was found that treating 1000 ischemic stroke and transient ischemic attack patient with high-dose atorvastatin for over one year will avert 4.8 ischemic strokes while causing 1.9 additional hemorrhagic strokes [11,12]. It is also interesting that post-stroke mortality is inversely related to cholesterol levels because higher cholesterol levels are associated with less severe strokes [13]. Finally, in a study of diabetics, while statin treatment reduced non-hemorrhagic strokes by 50%, the benefit was independent of baseline cholesterol and the presence or absence of a first stroke [14].

Taken together, all this evidence strongly suggests that the benefits of statins in the context of stroke has little to do with cholesterol lowering and may be due to other actions of these drugs.

LIPID LOWERING TRIALS
Today statins are with rare exceptions the only drugs given to lower cholesterol levels. Worldwide sales of 55 billion U.S. dollars suggest widespread use. It is widely accepted that the success of statin drugs in decreasing the risk of cardiovascular events provides convincing proof of the role of cholesterol in the etiology of CHD. As discussed in Part I of this review on cholesterol, this argument is seriously flawed. Since the use of statins is so widespread, it is of interest to examine the relative and absolute benefits associated the use of these drugs in both the primary and secondary setting and as well, the notion that when it comes to LDL, the lower the better and that almost everyone has elevated LDL and needs therapy.

Some lipid lowering trials were exclusively for primary prevention, some only for subjects with CHD, and some looked at both. There have been trials that exclusively enrolled men and those that had a mix of genders. In several studies, women have been seriously underrepresented. A number of different statins have been tested.

For primary prevention using statins the available studies are very limited. One is 100% male and the other two are 82-85% male. One [15] recruited only hypertensive individuals with at least three other CVD risk factors (ASCOT), and one [16] had a lower limit on baseline TC of 252 and a mean of 272 mg/dL, which translates into a significant percentage of individuals with family related high cholesterol (FH) (WOSCOPS). The one that came closest to reflecting characteristics of the general population was AFCAPS [17] where the mean TC was 221 and LDL ranged from 131 to 191, although this cohort had low HDL. Here are the results:

• WOSCOPS. 22% reduction in non-fatal CHD which was derived from a 1.9% absolute risk reduction. Thus the number needed to treat (NNT) to prevent one non-fatal CHD event was 53 and in the drug group 95.7% has no events compared to 93.8% in the placebo group during a 4.4 year follow-up. There was no significant benefit in terms of overall or CHD mortality. All the men had very high TC. The reduction of TC was 20% and LDL 26%.

• AFCAPS. The reduction in non-fatal CHD events was 38% but the absolute risk reduction was only 2%, yielding a NNT of 50 individuals. In the drug group, 96.5% experienced no adverse non-fatal CHD event whereas in the placebo group the number was 94.5%. There was no significant benefit in terms of overall mortality or CHD mortality. The cohort had average TC levels and was 85% male. The reduction of TC was 19% and LDL 26%.

• ASCOT. A 36% reduction in non-fatal heart attacks and fatal CHD was found. The absolute risk reduction was 1.1% giving a NNT of 91. No benefit was found in terms of overall mortality or cardiovascular mortality. In the treated group, 98.1% were free of non-fatal heart attacks or fatal CHD whereas for the placebo group the number was 97.0%. The cohort was comprised of hypertensive individuals with three other CHD risk factors and was 82% male. Reduction of TC was about 24% whereas LDL went down by about 32%.

A meta-analysis of seven primary prevention randomized controlled trials involving statins was published in 2006 which included the above three studies [18]. The results were as follows:

No significant effect of statin treatment on coronary heart disease mortality or overall mortality was found. For individuals in the Framingham low or intermediate risk categories, statin therapy was estimated to reduce the absolute risk of major coronary events by 0.75% and 1.63% respectively. Thus this analysis which involved almost 43,000 participants overall provides a similar picture to that for the three studies discussed in detail above. The absolute risk reduction is just slightly above negligible. However, from the public health point of view, even these small reductions would add up to a large number of events prevented in a large population such as the US, The benefit found for Framingham high-risk individuals was greater with a 2.51% absolute risk reduction, but the use of statins in this group is less contentious than for the low and intermediate risk groups.

It is not clear the extent to which these results apply to women since they were severely underrepresented. Also, the ASCOT trial does not apply to the general population, only a hypertensive population at high risk of CHD. Critics of the Cholesterol Hypothesis consider the small absolute risk reduction to be inconsistent with cholesterol being regarded as a large, significant and highly important risk factor. Believers say the above studies prove that serum cholesterol and in particular LDL actually causes atherosclerosis and CHD and in addition, lowering TC or LDL by 20-30% reduces CHD events by 20-30%. As discussed below, when the primary prevention studies are stratified for women of all ages and the elderly, the risk reduction of statin treatment disappears leaving only young men as beneficiaries, and some would regard number of young men needed to treat of 50-70 to prevent one adverse event as suggestive of marginal benefit. Treatment is generally for life.

Most of the research in lipid lowering has centered on secondary prevention and the comparison between low and high dose protocols. Regarding the former, Costa et al have provided a meta- analysis of 7 studies [19]. The result was as 23% reduction in major coronary events and the absolute risk reduction was 5.1% with the NNT about 20. In the meta-analysis discussed above [18], for trials where secondary prevention was also involved, the relative risk reduction for major coronary events was 20.8% with an absolute reduction of 2.4% and the number needed to treat of 33. In this study of secondary prevention trials there was also no statistical significant association between statin treatment and coronary heart disease mortality or overall mortality.

Studies aimed at evaluating the benefits of high-dose statins vs. the usual dose have examined either individuals with stable CHD or those experiencing an acute coronary syndrome. Cannon et al have performed a pooled analysis of 4 studies [8]. When the two dose protocols were compared, the reduction in risk of CHD mortality or heart attacks achieved for high dose statins was 16% with an absolute risk reduction of 1.4% and the NNT was 71. The benefits should be viewed as in addition to those obtained with the standard statin therapy. In the pooled analysis the LDL reductions were from 130 to 101 mg/dL for the usual dose and from 130 to 75 mg/dL for the high dose, so-called intensive treatment. No significant differences were found between low and high dose treatment for CVD mortality or overall mortality. The analysis involved over 100,000 patient years of observation.

This then is a snapshot of the research behind the statement that for secondary prevention, statins offer benefit in terms of preventing additional adverse coronary events. Critics of the Cholesterol Hypothesis do not appear to dispute this conclusion. Where they part company with the conventional wisdom concerns the use of these results as proof of the Cholesterol Hypothesis, frequently with statements embellished with superlatives. As will be discussed below, the use of high statin doses, presumably for life, raises questions about side effects over and above those already associated with the standard dose protocol.

PRIMARY PREVENTION: THE SPECIAL CASE OF WOMEN OF ALL AGES AND THE ELDERLY
In the January 20th 2007 issue of the journal Lancet, two researchers, one from Harvard (J. Abramson) the other from the University of British Columbia (J. M. Wright) raised serious questions about the extent of the evidence supporting the use of statin drugs for true primary prevention of cardiovascular events or life extension [20]. The authors acknowledge that for individuals between 30 and 80 years of age with occlusive vascular disease, secondary prevention with statins confers benefit. What is at issue here are individuals who exhibit no evidence of disease. They point out that about 75% of those taking statins are in this category, i.e. pure primary prevention. On the basis of analysis of pooled data published earlier [21] and as well, reference to specific studies, they conclude that there is no statistically significant evidence favoring the use of statins for pure primary prevention for the following subsets: (a) women of any age; (b) men older than 69 years. The authors claim that in justifying primary prevention with statins in women and in people over 65 years of age, the U.S. guidelines for treatment cite 16 randomized trials and yet not one provides evidence of benefit from statin therapy for these two groups. In addition, they find that high-risk men between 30 and 69 with no apparent vascular disease should be advised that about 50 patients need to be treated for 5 years to prevent one adverse event.

The pooled studies used by Abramson and Wright consisted of five large trials of statins including the three discussed above, which mostly involved primary prevention (average percent primary prevention—83% of participants, range 56-100%). In the pooled studies, total mortality was not reduced by statins and while the 5-year frequency of total heart attacks and stroke was reduced (relative risk 0.84) the absolute risk reduction was only 1.4%. This is equivalent to needing to treat 71 individuals for 5 years to prevent one event. They also quote the results of the PROSPER randomized controlled trial which involved over 5800 men and women over 69 years of age [22]. In a subset of 3239 men and women in this trial with no evidence of previous vascular disease and viewed at risk because of smoking, hypertension or diabetes, this study found that statins did not reduce total cardiovascular events. When the PROSPER results were stratified just by gender, among women there was no significant benefit from statin treatment and an unspecified number of the total female cohort actually had prior vascular disease. In the interpretation part of the abstract of the PROSPER paper no mention was made of the absence of benefit for primary prevention or for elderly women in general (primary or secondary prevention), but rather, it is simply stated that the statin in question given for 3 years reduced the risk of coronary disease in elderly individuals, thereby omitting an important result. This is in spite of the fact that of the 5804 individuals included in the analysis, about 56% were in the "no previous vascular disease" category, i.e. a significant fraction of the total study population. This data has been in the literature since 2002. Finally, the paper published in 2004 in the Journal of the American Medical Association to which Abramson and Wright make reference found for women without cardiovascular disease, cholesterol lowering with a statin drug did not affect total or coronary heart disease mortality. For fatal heart attacks, revascularization or coronary heart disease events, only one out of nine studies showed significant treatment benefit and this was just for one outcome of many reported in that study. Even for those with known cardiovascular disease, lipid lowering did not affect total mortality. This was a study of studies (meta-analysis) which included six trials involving 11435 women of various ages without cardiovascular disease [23]. Thus the evidence points to the conclusion that in the context of primary prevention for women of all ages and the elderly of either gender, cholesterol lowering is without benefit—a non- issue.

THE LOWER THE BETTER. WE ALL HAVE ELEVATED LDL
There is an interesting problem of why people with low as well as high LDL or TC get the same heart disease. One ad hoc hypothesis offered by those who defend the Cholesterol Hypothesis is that in fact in the Western world almost everyone has high cholesterol and therefore high LDL. The contention promoted by O'Keefe et al is that the optimal LDL for everyone is between 50 and 70 mg/dL [24]. For almost everyone, the only way to get LDL down to these levels would be lifelong use of drugs, currently statin drugs. This assertion regarding a universal syndrome of high LDL is based on the observation that the LDL range is 50-70 mg/dL in native hunter-gatherers, newborns, free-living primates and wild animals such as the baboon, howler monkey, horse, black rhinoceros, and African elephant. However, the hunter-gather societies looked at by the proponents of this theory all had very short life expectancies ranging from 17 years to 36 years. These individuals are hardly comparable to those residing in the developed part of the world. Also, it is debatable if newborns are a good standard. If we attempted to achieve the low blood pressures commonly seen in newborn children, we would probably in the process kill ourselves. The relevance of cholesterol levels in animals ranging in size from small to huge is also not obvious, especially since they are not carnivores. Thus there appear to be problems with this theory. In this context it is interesting that in Japan when cholesterol levels rose from 150 to 190, life expectancy increased and CVD fell dramatically. Also, as discussed at length in Part II, Cholesterol vs. mortality studies consistently show a J-shaped curve for men with mortality increasing as TC or LDL decrease to low values and a flat dependence for women as TC is decreased from high values until at low values the mortality also rises like that seen for men. Most of these studies are designed so that concurrent disease that might lower cholesterol is excluded. The 30-year report from the Framingham study also found that when cholesterol declined by itself rather than through the action of drugs, mortality increased rather than decreased. Thus even if one is convinced by blood cholesterol studies on new-born babies, modern-day hunter-gatherers and a marvellous assortment of animals, they still have to deal with this mortality issue in adult humans.

O'Keefe et al also attempt to justify their belief that every one should have an LDL level between 50 and 70 mg/dL by presenting some diagrams of CHD events vs. LDL and showing a linear correlations where CHD events go to zero in this LDL range. However, there are problems with this approach. Let's consider the correlation they obtained based on primary prevention trial data. Three studies were used. One involved only men with very high LDL and TC and one involved only hypertensives with three additional risk factors. The third looked at mostly men with average cholesterol levels. When the CHD event rates are plotted vs. the baseline average LDL and the average LDL achieved after statin treatment, a straight line results going to zero events at an LDL of 55 mg/dL. But these three populations are not comparable and this approach involved no corrections for confounding. The high cholesterol group probably included individuals with FH. And data from only three studies were used. If one wants to examine the correlation between cholesterol and CHD events, this highly limited data set uncorrected for confounding does not appear to be a very good way to do it. And this analysis also ignores the great danger of extrapolating data. Finally, the many results discussed in Part I where no correlation was found among asymptomatic individuals between CAD events and TC (and thus LDL) except for young men argues against the validity of this or the other similar correlations used by O'Keefe et al.

The acceptance and implementation of the notion that only LDL in the range of 50-70 mg/dL is healthy and that if one does not meet this criterion, lipid lowering is in order, would put a significant fraction of the developed world and the rich and well-off elsewhere on statins for life since there is currently no other means of achieving these very low levels. Exercise and diet would for most not accomplish such reductions, given that such lifestyle interventions are already well known to produce only small changes in TC or LDL.

SIDE EFFECTS WITH STATIN THERAPY
It appears that mainstream medicine is moving toward lower and lower targets for optimal LDL and, as mentioned above is flirting with the notion that it should be as low or even lower than 70 mg/dL for everyone. What appear to be limited benefits only for a very small sup-population in the developed world and as well data suggesting higher mortality at low LDL levels seem to be ignored. But this is part of a larger question—the side effects in general of statin drugs. This is an issue when any risk-benefit analysis, formal or informal, is undertaken. What are the side effects, how serious are they, and are they significantly underreported and downplayed?

Before a drug is approved, drug companies must study adverse side effects. This is generally done with at most several thousand subjects over a fairly short period, a system that is bound to fail. If we had mandatory long-term (say 5-10 years) testing for side effects in a cohort of say 10 or 20 thousand subjects, the introduction of new drugs would decline precipitously. In fact, the industry would probably never allow this to happen. The point is that only the most prevalent side effects manage to surface. After the drug is approved, a mechanism exists for the reporting of adverse side effects, but it is well known that the compliance is negligible with only about 1% of the cases reported. When very serious side effects were just below the threshold of detectability during clinical trials but surfaced when several million prescriptions are written, it is not uncommon that the drug is recalled, although the period of denial can be considerable (e.g. the Vioxx incident). Side effects either ignored or missed in the clinical trials have to kill or disable enough patients for someone to notice. Numerous examples could be quoted. With the statin drugs, it is very easy for a physician to regard side effects as simply manifestations of aging or related to some comorbidity or lifestyle excess. There must also be an element of denial involved since statins are taken by millions worldwide every day, so how can there be any serious problems? It is interesting that in veterinarian medicine both alertness to adverse effects and a willingness to immediately take action appears much greater than in human medicine. Evidence—the very rapid recognition and equally rapid response to the recent contaminated pet food incident.

Comprehensive reviews of the side effects of statin drugs are mostly to be found outside mainstream medical literature and are thus easily dismissed or ignored by the establishment. But some of the voices have considerable credibility. For example, the author of Statin Drugs-- Side Effects and the Misguided War on Cholesterol and Lipitor--Thief of Memory is Dr. Duane Graveline, M.D., a former astronaut, an aerospace medical research scientist, NASA flight surgeon. Another voice is that of Dr. Mary G. Enig, Ph.D., a well-known and respected authority on lipid biochemistry and the author of over 60 publications and presentations. She is the author of Know your Fats: The complete Primer for Understanding the Nutrition of Fats, Oils and Cholesterol.

Graveline's most recent book on Lipitor and memory provides an account of his shocking experiences with what is called transient global amnesia (TGA), a sudden temporary total loss of memory where one does not recognize anything or anybody and fails to have memory of past events going back decades. When his first incident occurred, no one was able identify a cause. After he recovered he reasoned that it might be medication related. He had been put on Lipitor and immediately discontinued it. No more episodes occurred until he resumed Lipitor under pressure from his physician. Then he had a second and more serious episode. This prompted him to set up a website and collect information on statin side effects and in particular TGA. Over several years he has collected enough data to convince himself that TGA is much more common than acknowledged. This is not a trivial matter since there are many occupations where such an episode could prove dangerous to others than the victim, e.g. an airline pilot at the controls during a final approach. In his book on side effects Graveline provides a comprehensive review which should be of interest to anyone taking or contemplating taking statins. His website www.spacedoc.net updates his continuing research and data collection on adverse side effects and in particular what is reported to the government site (Medwatch).

Mary Enig and Sally Fallon have written a review of statin side effects which was published by the Weston A. Price foundation and can be found on its website www.westonaprice.org. They point out what most scientists involved in cholesterol research know, statins not only interfere with cholesterol synthesis but also with a number of other biochemical pathways which generate substances of critical biological importance. The two most important are co-enzyme Q10, a critical cellular micronutrient present in all mitochondria, and a family of compounds call dolichols which in cell biochemistry direct manufactured proteins to their proper targets in response to DNA directives, ensuring that cells respond correctly to genetically programmed instructions. Also, squalene, an intermediate in the reaction chain that yields cholesterol, has anti-cancer properties.

But statins also reduce the synthesis of another compound, mevalonate, which among other things is involved in clotting mechanisms and thus might cause changes that are in a favorable direction. There are other potentially beneficial biochemical actions of statins that are under intense investigation by scientists impressed by the growing evidence that statins act by mechanisms independent of cholesterol lowering. Lipid lowering is also very effective in relieving anxiety about elevated cholesterol levels, and this may have significant health benefits.

But to return to adverse side effects, let it suffice to list them: (a) muscle pain and weakness including potentially fatal rhabdomyolysis; (b) peripheral neuropathy and nerve damage which can be permanent; (c) heart failure thought to be partly due to a deficiency of co-enzyme Q-10, a biochemical that incidentally is used in some countries as a prescription drug to treat heart failure; (d) dizziness and large drops in blood pressure; (e) cancer and the depression of the immune system. Statins have even been suggested for use with transplant patients as an anti-rejection drug; (f) depression linked to low cholesterol levels. This list comes from the review by Enig and Fallon cited above. More details can be found in Graveline's books, which also include some interesting case histories. It is probably safe to say nobody really knows the prevalence of the adverse side effects of this class of drug. This is also not a popular or prestigious area of research, nor one where it is easy to get funding.

In his book on side effects cited above, Graveline presents numerous anecdotal reports where the individuals involved though their problems were connected to statins, but their symptoms were attributed to Alzheimer's disease, Parkinson's disease, amyotrophic lateral sclerosis, i.e. (Lou Gehrig's disease (ALS), or multiple sclerosis. This puts the problem in clear perspective. The incidence of such disorders increases with age and there are no widely recognized or accepted studies that directly relate statin use to these symptoms. Thus while physicians would not be surprised to see an increase in patients presenting with symptoms of these neurological disorders, in general they would not make any association with statins. Statins after all, except for what the industry describe as very rare muscle problems, are thought to be safe. The point is, while it may be true that statin users are putting themselves at increased risk of neurological disorders which may mimic the diseases listed above, it is unlikely that mainstream medicine will recognize the potential connection. The association if present may never be established to the satisfaction of those demanding evidence. The studies are too long, too difficult to organize, and too difficult to interpret, given that the incidence of the problems in question is rare, although certainly significant to those who experience the symptoms. And who is going to pay for them? What is really alarming about the anecdotal reports given by Graveline is that in most cases, while symptoms diminished after termination of statin use, in some individuals they did not completely resolve, leaving patients worried that the damage sustained, possibly by statin use, was permanent. No one appears to be seriously studying these issues. In the real world of evidence based medicine, anecdotal reports are easily dismissed. Finally, in this review evidence was presented that cholesterol was not an issue for women or the elderly provided they did not have CHD. Yet there are millions of women of all ages and as well elderly individuals who are on statins and yet to not have CHD. If indeed they do not need statins, then they are unnecessarily putting themselves at risk for serious side effects.

This is not just an issue that cholesterol critics have emphasized. There is a class action lawsuit before the courts in the US claiming damages from a drug company for encouraging the medical profession to prescribe statins to women and the elderly. According to the submission to the court, the case is based on the contention that there is no evidence from relevant studies indicating benefit and there is a risk of harm for the age-gender groups in question.

These then are some of the considerations that come into play when trying to weigh the risks and benefits of statin therapy. If one believes that cholesterol is not an issue for women or for the elderly who do not have CHD, then one side of the scale has nothing on it. For young men, this is an important issue given the large number of individuals needed to treat to prevent one adverse CHD incident.

STATINS FOR CHILDREN ???
The American Heart Association has just published its latest guidelines regarding drug therapy for high-risk lipid abnormalities in children and adolescents [25]. These represent modifications of the 1992 National Cholesterol Education Program guidelines. The main features of the older guidelines have been preserved, and merely modified on the basis of what the committee regards as the current situation with regard to other risk factors that might influence the decision for drug intervention. Thus the current guidelines are as follows.

1. Consider drug intervention in children age greater than equal to 10 (usually wait until menarche for females and after a 6-12 month trial of diet and exercise).
2. Consider drug therapy if LDL remains greater than equal to 190 mg/dL (4.90 mmol/L). If LDL remains > 160 mg/mL (4.1 mmole/LL) and (a) there is a positive family history of premature cardiovascular disease or (b) greater than equal to 2 other risk factors are present after vigorous attempts to control these factors, then drug intervention is also indicated.
3. Treatment goal: Minimal, LDL < 130 mg/mL (3.35 mmole/LL; ideal < 110 mg/mL (2.85 mmole/LL)

Risk factors and high-risk conditions: Male gender, low HDL, high triglycerides, small dense LDL, overweight or obese and other aspects of the metabolic syndrome, hypertension, smoking or exposure to passive smoke, and elevated lipoprotein(a), homocysteine, or c-reactive protein. Finally, there are medical conditions that increase the risk such as diabetes, HIV infection, systemic lupus erythematosus, organ transplantation or surviving childhood cancer and these also count as risk factors.

The authors comment that the evidence base underlying these recommendations including LDL targets is supported only by indirect evidence, evidence extrapolated from studies on adults, studies performed in the context of familial hypercholesterolemia and expert consensus. Furthermore, they caution that direct evidence of an impact of interventions during childhood and adolescence on later cardiovascular morbidity and mortality will likely always be lacking and that drug therapy should only be targeted toward individuals with high-risk lipid abnormalities or high- risk conditions who have not reached the target lipid levels with lifestyle modification.

First, how common are levels of 190 or 160 mg/dL? A recent study looked at averages from data collected between 1988 and 2002 [26]. For the age group from 12 to 20 years, mean LDL values were about 100 mg/dL. For men, the 98th percentile came in at about 155 mg/dL and for women the 95th percentile was about 135 mg/dL. Thus for this age group, LDL > 160 or especially = 190 is rare and appears to truly represent an abnormality. But In the group of children with LDL > than 160 mg/dL, if only two of the above listed risk factors are necessary for triggering drug intervention, then this no doubt expands the eligible group considerably. Second, how successful are dietary interventions in correcting elevated LDL. In a study published in 2001, a conventional intervention involving reductions in total fat, saturated fat and dietary cholesterol over 5-8 years produced a 2.0 mg/dL drop in LDL when comparison was made with a control group. Furthermore, for this group which was initially 8-10 years old, both the intervention and control groups experienced a drop in LDL from about 130 to 110 mg/dL on 5-8 years follow-up [27]. Obviously, a 2 mg/dL decrease in LDL would seem to have little clinical significance, especially if one has an LDL over 160 and measurements in this age group may have little value anyway.

However, there are fundamental questions. We are dealing here with recommendations that are by and large not based on studies relevant to the age group in question nor to the intended long- term use of drug intervention. In a special article for the journal Pediatrics, Belay et al address these and related issues [28]. They make the following points: (a) Randomized clinical trials of statins in adults have been mostly for secondary prevention in older adults, and because even children considered at the highest risk rarely experience a cardiovascular event, the results of secondary prevention trials in adults should be cautiously applied to children; (b) the translation of aggressive treatment targets based on secondary prevention trials from adults to children and adolescents is not justified for primary prevention; (c) primary prevention studies using statins with adults have not demonstrated any reduction in absolute risk for total mortality. They regard this as a crucial aspect since before statins are used for primary prevention in children and adolescents, they consider it critical to demonstrate that total mortality is reduced in later life with childhood statin therapy; (d) the anti-inflammatory benefits of statins on advanced atherosclerotic plaques may not be seen in children whose plaques are at a much different stage of development. They go so far as to suggest that the effect of statins seen in primary prevention relates to the subgroups of individuals who already have unstable plaques; (e) with regard to the safety issue, the AHA position which incidentally seems to downplay side effects, is based on trials with children that have lasted only from 6 months to 2 years and the clinical trials on adolescents and children have been underpowered (too few subjects) to detect infrequent or rare adverse effects. They also point to the problem that the intervention is during cognitive and endocrinologic maturation, skeletal growth, and bone mineral accretion and that there are no studies that directly address other statin safety issues in this age group.

Other problems need to be mentioned. Statins may be associated with central nervous system problems and limb anomalies in about 15% of exposed first-trimester pregnancies [29]. Therefore, statin therapy among female teenagers capable of reproduction may carry a very significant birth defect danger unless contraceptives are successfully employed. Also, girls have lower risk of developing cardiovascular disease than boys, and this must be taken into account in decisions about who to treat. There are also issues associated with breast cancer risk associated with use of oral contraceptives prior to about age 20 (see the recent review on primary prevention of breast cancer in the archives of IHN).

Finally, a study published in 1990 looked at cholesterol levels in children 8 to 18 years of age and then followed them for 20-30 years to see how many as adults developed cholesterol levels that would have merited continued surveillance and intervention [30]. The researchers found that screening for total cholesterol resulted in significant numbers of individuals being incorrectly classified with respect to future cholesterol level elevations. They point out that based on their results, many children with high cholesterol levels have normal levels in young adulthood without intervention.

While the above considerations seem to make it clear that the AHA recommendations are not, as the authors admit, really evidence based and that there are many important issues, it is however almost impossible to implement studies that start with very young children, put half of them on statins, and follow the cohort for 30-40 years to see if the childhood LDL levels and risk factors that would trigger this drug intervention in childhood really lead to cardiovascular problems in much later life which are significantly decreased in the treatment group. Early treatment vs. treatment of adult disease has never been carefully investigated. Thus the bottom line in the case of children deemed at risk appears to be very aggressive lifestyle intervention to correct childhood obesity and poor dietary habits. But the guidelines talk about diets low in cholesterol and saturated fat, and yet the studies they quote found this intervention had almost no effect on LDL levels. Other studies cited in the guidelines paper also were impractical (no meat or dairy products) or produced only small effects and these on children with normal LDL levels. They also are inconsistent in that they reference the AHA pediatric diet strategies which do not mention avoiding cholesterol containing foods and which appears more in tune with reality [31]. To get the diet plus exercise approach to really work will, it would seem, present a severe but obviously highly worthwhile challenge to parents with children who might be targeted for statin therapy.

In addition, guidelines, which severely restrict fat, will in this age group probably invariably result in a large increase in refined carbohydrate intake with an associated increase in triglycerides and HDL, an increase in the risk of insulin resistance and inflammation, and thus an increase in the risk of atherosclerosis. Just what is being targeted for prevention.

Given that (a) the indications for statin use in children are very far from securely evidence based and in fact mostly based on evidence from adult studies on individuals with heart disease; (b) that we are presumably talking about very long-term therapy; (c) that the long-term safety has not and may never be directly established for this age group; and (d) that the optimum age for pharmaceutical intervention has not been established, it would seem that parents need to be very concerned regarding the recommendation to proceed with childhood statin therapy. But it must also be acknowledged that children with a strong family history of premature heart disease who have very high cholesterol levels are a special case for which some of the objections and cautions enumerated above may perhaps assume somewhat less important. But this does not minimize the problem of the absence of relevant evidence of benefit in this age group.

STATIN BENEFITS UNRELATED TO CHOLESTEROL LOWERING
In a recent review, Ray et al summarize the evidence suggesting that some of the benefits of statin drugs are independent of lipid lowering [32]. They include

• Benefits appear independent of baseline LDL level.
• Benefits exceed the benefit predicted by the change in LDL level.
• Rapid benefits of aggressive initiation of statin therapy which come long before there is a change in cholesterol levels.
• The different efficacy between statins is unrelated to their effects on cholesterol levels.

Studies have revealed that statin-mediated improvements in endothelial function (as for example measured by improved blood flow) occur within days in humans, even after a single dose and before any significant effects on lipids [33]. In fact, a recent study found that a non-statin cholesterol lowering drug (ezetimibe) which was effective in lowering LDL levels failed to improve endothelial function whereas equivalent lipid lowering with a statin produced the benefit [33]. These are called pleiotropic functions of a therapy, and at present this is a very active and expanding area of research.

These pleiotropic functions of statin drugs undermine the argument that is the cornerstone of the Cholesterol Hypothesis. This now appears to be gaining some recognition. For example, Brotman et al [34] state in a recent article that "observing that statins reduce LDL cholesterol levels while reducing cardiovascular mortality does not prove that LDL causes CHD, since statins also affect other cardiovascular risk factors." They continue, "Since statins affect inflammation, endothelial function, oxidative stress, and coagulation, we cannot conclude that LDL cholesterol is atherogenic based on statin studies alone. This requires as convincing mechanistic explanation and an array of consistent evidence supporting it." Also, in an editorial regarding the SPARCL study [9] which found 80 mg/day of atorvastatin reduced overall incidence of stroke and cardiovascular events, David Kent remarks that [10] "However, the finding has not settled the controversy [cholesterol and stroke], since a strong correlation between the reduction of cholesterol level and stroke prevention would be expected regardless of the mechanism by which statins decrease risk, i.e. regardless of whether the cholesterol level is the actual mediator of the treatment effect or merely a marker of adequate therapy and adherence."

SUMMARY
Ignoring the lipid lowering results, which we have dismissed as not proving the Hypothesis, the evidence that high total cholesterol or LDL cholesterol causes atherosclerosis and CHD appears limited to men under 50. An explanation for this exception was presented based on the hypothesis that this group may have on average a very high exposure to professional and domestic stress. In this age group there will be individuals who have exaggerated blood pressure response to stress and this is associated with high levels of cholesterol. Exaggerated blood pressure response is itself a risk factor for CHD, and thus the correlation with cholesterol. TC and LDL are bystanders, not causative agents in this view [35]. The arguments that family- related high cholesterol (FH) proves the Cholesterol Hypothesis is falsified if one accepts that atherosclerosis is different in young FH individuals, that FH has a number of other adverse aspects, and that in general, there is no evidence that FH on average reduces life expectancy. One has also to consider the large number of observations discussed in this review that tend to falsify the Cholesterol Hypothesis or weaken or destroy its foundations. In four books on this subject published since 2001, three by medical doctors, the phrase "The Cholesterol Myth" has set the tone. All four books essentially tell the same story as elaborated in this review. One has a nice cartoon showing spectators at a parade remarking the Cholesterol Hypothesis Emperor seated on his throne in a float has no clothes. But the Hypothesis is alive, well, thriving and in fact gaining momentum as more and more people become eligible for lipid lowering treatment under frequently revised guidelines, and this includes women of all ages and the elderly, both with no evidence of CHD, for whom there appears to be no significant evidence of benefit. Finally we have the now famous proposal of Wald and Law that everyone over that age of 55 should take the so-called Polypill since, in their opinion, it could largely prevent heart attacks and stroke [36]. This is not a joke but a serious proposal published in the British Medical Journal, a peer reviewed publication. The Polypill contains a statin, low-dose aspirin and a high blood pressure drug and would presumably be patentable. In the over-medicated developed world, this actually seems like a natural endpoint in the evolution of we have been witnessing in mainstream medicine's approach to health and disease.

This review has more or less ignored HDL cholesterol and the evidence that high levels are indeed protective. The subject of HDL is closely tied to triglycerides and to diet. A diet high in carbohydrates, and especially one with lots of refined carbohydrates such as refined starches and sugar, will in many individuals dramatically raise triglyceride levels and significantly lower HDL levels and at the same time increase the risk of developing insulin resistance and systemic inflammation which includes inflammation of arteries. This places one on the road to atherosclerosis, diabetes, obesity and CHD. These adverse blood lipid changes are accompanied by a shift in the LDL particle size in the direction of small, dense LDL particles thought to be the important fraction in the context of inflammation, atherosclerosis and CHD. In fact, high triglycerides and low HDL are considered to be a good surrogate marker for insulin resistance, and insulin resistance is definitely something to be avoided. Nevertheless, current guidelines include recommendations for low-fat diets and in some cases very low-fat diets and many individuals would find it very difficult to make up the calorie deficit with fruit and green vegetables and a minimum of protein, the latter being avoided because of the fat that many protein sources contain. They turn of necessity to refined carbohydrates in large amounts and thereby defeat the purpose of the dietary recommendation, i.e. reducing the risk of CVD, by raising their triglycerides, lowering their HDL and ultimately acquiring insulin resistance, changes which work in exactly the opposite direction from the goals envisioned by the guidelines.

REFERENCES - PART I

1. Miida T, Takahashi A, Ikeuchi T, 2007. Prevention of stroke and dementia by statin therapy: experimental and clinical evidence of their pleiotropic effects. Pharmacol.Ther. 113(2):378-393.
2. Liao JK, 2002. Beyond lipid lowering: the role of statins in vascular protection. Int.J Cardiol. 86(1):5- 18.
3. Kendrick M, The Great Cholesterol Con. The Truth About What Really Causes Heart Disease and How to Avoid It. John Blake Publishing Ltd., London, U.K., 2007.
4. Lande K, Sperry W, 1936. Human Atherosclerosis in Relation to the Cholesterol Content of the Blood Serum. Archives of Pathology. 22:301-312.
5. Paterson J, Armstrong R, Armstrong E, 1963. Serum lipid levels and the severity of coronary and cerebral atherosclerosis in adequately nourished men. Circulation. 27:229-236.
6. Mathur KS et al., 1961. Serum cholesterol and atherosclerosis in man. Circulation. 23:847-852.
7. Ravnskov U, 2002. Is atherosclerosis caused by high cholesterol? QJM. 95(6):397-403.
8. Ravnskov U, The Cholesterol Myths. Exposing the Fallacy that Saturated Fat and Cholesterol Cause Heart Disease. NewTrends Publishing, Inc, Washington, D.C., 2000.
9. Kondos GT et al., 2003. Electron-beam tomography coronary artery calcium and cardiac events: a 37-month follow-up of 5635 initially asymptomatic low- to intermediate-risk adults. Circulation. 107(20):2571-2576.
10. Allison MA, Wright CM, 2005. Age and gender are the strongest clinical correlates of prevalent coronary calcification (R1). Int.J Cardiol. 98(2):325-330.
11. Oei HH et al., 2004. Risk factors for coronary calcification in older subjects. The Rotterdam Coronary Calcification Study. Eur.Heart J. 25(1):48-55.
12. Sekikawa A et al., 2005. Much lower prevalence of coronary calcium detected by electron-beam computed tomography among men aged 40-49 in Japan than in the US, despite a less favorable profile of major risk factors. Int.J Epidemiol. 34(1):173-179.
13. Nasir K et al., 2006. Relationship of subclinical coronary atherosclerosis and National Cholesterol Education Panel guidelines in asymptomatic Brazilian men. Int.J Cardiol. 108(1):68-75.
14. Aiyer AN et al., 2007. Racial differences in coronary artery calcification are not attributed to differences in lipoprotein particle sizes: the Heart Strategies Concentrating on Risk Evaluation (Heart SCORE) Study. Am.Heart J. 153(2):328-334.
15. Hoff JA et al., 2003. Conventional coronary artery disease risk factors and coronary artery calcium detected by electron beam tomography in 30,908 healthy individuals. Ann Epidemiol. 13(3):163-169.
16. Arad Y et al., 2005. Coronary calcification, coronary disease risk factors, C-reactive protein, and atherosclerotic cardiovascular disease events: the St. Francis Heart Study. J Am.Coll.Cardiol. 46(1):158-165.
17. Hecht HS et al., 2001. Relation of coronary artery calcium identified by electron beam tomography to serum lipoprotein levels and implications for treatment. Am J Cardiol. 87(4):406-412.
18. Kronmal RA et al., 2007. Risk factors for the progression of coronary artery calcification in asymptomatic subjects: results from the Multi-Ethnic Study of Atherosclerosis (MESA). Circulation. 115(21):2722-2730.
19. Ravnskov U, 2003. High cholesterol may protect against infections and atherosclerosis. QJM. 96(12):927-934.
20. Marmot MG et al., 1975. Epidemiologic studies of coronary heart disease and stroke in Japanese men living in Japan, Hawaii and California: prevalence of coronary and hypertensive heart disease and associated risk factors. Am.J Epidemiol. 102(6):514-525.
21. Marmot MG, Syme SL, 1976. Acculturation and coronary heart disease in Japanese-Americans. Am.J Epidemiol. 104(3):225-247.
22. Anderson KM, Castelli WP, Levy D, 1987. Cholesterol and mortality. 30 years of follow-up from the Framingham study. JAMA. 257(16):2176-2180.
23. Allison MA, Wright CM, 2005. Age and gender are the strongest clinical correlates of prevalent coronary calcification (R1). Int.J Cardiol. 98(2):325-330.
24. Greenland P et al., 2004. Coronary artery calcium score combined with Framingham score for risk prediction in asymptomatic individuals. JAMA. 291(2):210-215.
25. Desai MY et al., 2004. Underlying risk factors incrementally add to the standard risk estimate in detecting subclinical atherosclerosis in low- and intermediate-risk middle-aged asymptomatic individuals. Am Heart J. 148(5):871-877.
26. Iribarren C et al., 2000. Association of hostility with coronary artery calcification in young adults: the CARDIA study. Coronary Artery Risk Development in Young Adults. JAMA. 283(19):2546-2551.
27. Agatisa PK et al., 2005. Coronary and aortic calcification in women with a history of major depression. Arch Intern Med. 165(11):1229-1236.
28. Matthews K et al., 2006. Diurnal cortisol decline is related to coronary calcification: CARDIA study. Psychosom.Med. 68(5):657-661.
29. Rosengren A et al., 2004. Association of psychosocial risk factors with risk of acute myocardial infarction in 11119 cases and 13648 controls from 52 countries (the INTERHEART study): case-control study. Lancet. 364(9438):953-962.
30. Yusuf S et al., 2004. Effect of potentially modifiable risk factors associated with myocardial infarction in 52 countries (the INTERHEART study): case-control study. Lancet. 364(9438):937-952.

REFERENCES - PART II

1. Ravnskov U, The Cholesterol Myths. Exposing the Fallacy that Saturated Fat and Cholesterol Cause Heart Disease. NewTrends Publishing, Inc, Washington, D.C., 2000.
2. Kendrick M, The Great Cholesterol Con. The Truth About What Really Causes Heart Disease and How to Avoid It. John Blake Publishing Ltd., London, U.K., 2007.
3. de Waard F, Thijssen JH, 2005. Hormonal aspects in the causation of human breast cancer: epidemiological hypotheses reviewed, with special reference to nutritional status and first pregnancy. J Steroid Biochem.Mol.Biol. 97(5):451-458.
4. Ulmer H et al., 2004. Why Eve is not Adam: prospective follow-up in 149650 women and men of cholesterol and other risk factors related to cardiovascular and all-cause mortality. J Womens Health (Larchmt.). 13(1):41-53.
5. Ravnskov U, 2003. High cholesterol may protect against infections and atherosclerosis. QJM. 96(12):927-934.
6. Schatz IJ et al., 2001. Cholesterol and all-cause mortality in elderly people from the Honolulu Heart Program: a cohort study. Lancet. 358(9279):351-355.
7. Lowe LP et al., 1998. Impact of major cardiovascular disease risk factors, particularly in combination, on 22-year mortality in women and men. Arch Intern Med. 158(18):2007-2014.
8. Song YM, Sung J, Kim JS, 2000. Which cholesterol level is related to the lowest mortality in a population with low mean cholesterol level: a 6.4-year follow-up study of 482,472 Korean men. Am.J Epidemiol. 151(8):739-747.
9. Spada RS et al., 2007. Low total cholesterol predicts mortality in the nondemented oldest old. Arch Gerontol.Geriatr. 44 Suppl 1:381-384.
10. Choi JS, Song YM, Sung J, 2007. Serum total cholesterol and mortality in middle-aged Korean women. Atherosclerosis. 192(2):445-447.
11. Olsen TS et al., 2007. Higher total serum cholesterol levels are associated with less severe strokes and lower all-cause mortality: ten-year follow-up of ischemic strokes in the Copenhagen Stroke Study. Stroke. 38(10):2646-2651.
12. Stamler J et al., 2000. Relationship of baseline serum cholesterol levels in 3 large cohorts of younger men to long-term coronary, cardiovascular, and all-cause mortality and to longevity. JAMA. 284(3):311-318.
13. de Lorgeril M, Salen P, 2006. Cholesterol lowering and mortality: time for a new paradigm? Nutr.Metab Cardiovasc.Dis. 16(6):387-390.
14. Albert CM et al., 2003. Prospective study of sudden cardiac death among women in the United States. Circulation. 107(16):2096-2101.
15. Albert CM et al., 2002. Prospective study of C-reactive protein, homocysteine, and plasma lipid levels as predictors of sudden cardiac death. Circulation. 105(22):2595-2599.
16. 2001. Executive Summary of The Third Report of The National Cholesterol Education Program (NCEP) Expert Panel on Detection, Evaluation, And Treatment of High Blood Cholesterol In Adults (Adult Treatment Panel III). JAMA. 285(19):2486-2497.
17. Ware WR, 2007. High cholesterol and coronary heart disease in younger men: The potential role of stress induced exaggerated blood pressure response. Med Hypotheses.: e-published ahead of print.
18. Yusuf S et al., 2004. Effect of potentially modifiable risk factors associated with myocardial infarction in 52 countries (the INTERHEART study): case-control study. Lancet. 364(9438):937-952.
19. Jousilahti P et al., 1999. Sex, age, cardiovascular risk factors, and coronary heart disease: a prospective follow-up study of 14 786 middle-aged men and women in Finland. Circulation. 99(9):1165- 1172.
20. Bruno G et al., 2006. Effect of age on the association of non-high-density-lipoprotein cholesterol and apolipoprotein B with cardiovascular mortality in a Mediterranean population with type 2 diabetes: the Casale Monferrato study. Diabetologia. 49(5):937-944.

REFERENCES - PART III

1. 1995. Cholesterol, diastolic blood pressure, and stroke: 13,000 strokes in 450,000 people in 45 prospective cohorts. Prospective studies collaboration. Lancet. 346(8991-8992):1647-1653.
2. Law MR, Wald NJ, Rudnicka AR, 2003. Quantifying effect of statins on low density lipoprotein cholesterol, ischaemic heart disease, and stroke: systematic review and meta-analysis. BMJ. 326(7404):1423.
3. Willett W, Nutritional Epidemiology. Oxford University Press, New York, 1998.
4. Bots ML et al., 2002. Total and HDL cholesterol and risk of stroke. EUROSTROKE: a collaborative study among research centres in Europe. J Epidemiol Community Health. 56 Suppl 1:i19- i24.
5. Ebrahim S et al., 2006. Serum cholesterol, haemorrhagic stroke, ischaemic stroke, and myocardial infarction: Korean national health system prospective cohort study. BMJ. 333(7557):22.
6. Endres M, 2005. Statins and stroke. J Cereb.Blood Flow Metab. 25(9):1093-1110.
7. Amarenco P, Tonkin AM, 2004. Statins for stroke prevention: disappointment and hope. Circulation. 109(23 Suppl 1):III44-III49.
8. Cannon CP et al., 2006. Meta-analysis of cardiovascular outcomes trials comparing intensive versus moderate statin therapy. J Am.Coll.Cardiol. 48(3):438-445.
9. Amarenco P et al., 2006. High-dose atorvastatin after stroke or transient ischemic attack. The New England Journal of Medicine. 355(6):549-559.
10. Kent DM, 2006. Stroke--an equal opportunity for the initiation of statin therapy. The New England Journal of Medicine. 355(6):613-615.
11. Ovbiagele B, Saver JL, 2007. Intensive statin therapy after stroke or transient ischemic attack: a SPARCLing success? Stroke. 38(3):1110-1112.
12. Amarenco P et al., 2007. Effects of Intense Low-Density Lipoprotein Cholesterol Reduction in Patients With Stroke or Transient Ischemic Attack. The Stroke Prevention by Aggressive Reduction in Cholesterol Levels (SPARCL) Trial. Stroke.
13. Olsen TS et al., 2007. Higher total serum cholesterol levels are associated with less severe strokes and lower all-cause mortality: ten-year follow-up of ischemic strokes in the Copenhagen Stroke Study. Stroke. 38(10):2646-2651.
14. Hitman GA et al., 2007. Stroke prediction and stroke prevention with atorvastatin in the Collaborative Atorvastatin Diabetes Study (CARDS). Diabet.Med.
15. Sever PS et al., 2003. Prevention of coronary and stroke events with atorvastatin in hypertensive patients who have average or lower-than-average cholesterol concentrations, in the Anglo- Scandinavian Cardiac Outcomes Trial--Lipid Lowering Arm (ASCOT-LLA): a multicentre randomised controlled trial. Lancet. 361(9364):1149-1158.
16. Shepherd J et al., 1995. Prevention of coronary heart disease with pravastatin in men with hypercholesterolemia. West of Scotland Coronary Prevention Study Group. The New England Journal of Medicine. 333(20):1301-1307.
17. Downs JR et al., 1998. Primary prevention of acute coronary events with lovastatin in men and women with average cholesterol levels: results of AFCAPS/TexCAPS. Air Force/Texas Coronary Atherosclerosis Prevention Study. JAMA. 279(20):1615-1622.
18. Thavendiranathan P et al., 2006. Primary prevention of cardiovascular diseases with statin therapy: a meta-analysis of randomized controlled trials. Arch Intern Med. 166(21):2307-2313.
19. Costa J et al., 2006. Efficacy of lipid lowering drug treatment for diabetic and non-diabetic patients: meta-analysis of randomised controlled trials. BMJ. 332(7550):1115-1124.
20. Abramson J, Wright JM, 2007. Are lipid-lowering guidelines evidence-based? Lancet. 369(9557):168-169.
21. Jauca CaWJM, 2003. Update on Statin Therapy. Int Soc Drug Bull Newsletter. 17(3):7-9.
22. Shepherd J et al., 2002. Pravastatin in elderly individuals at risk of vascular disease (PROSPER): a randomised controlled trial. Lancet. 360(9346):1623-1630.
23. Walsh JM, Pignone M, 2004. Drug treatment of hyperlipidemia in women. JAMA. 291(18):2243- 2252.
24. O'Keefe JH, Jr. et al., 2004. Optimal low-density lipoprotein is 50 to 70 mg/dl: lower is better and physiologically normal. J Am.Coll.Cardiol. 43(11):2142-2146.
25. McCrindle BW et al., 2007. Drug therapy of high-risk lipid abnormalities in children and adolescents: a scientific statement from the American Heart Association Atherosclerosis, Hypertension, and Obesity in Youth Committee, Council of Cardiovascular Disease in the Young, with the Council on Cardiovascular Nursing. Circulation. 115(14):1948-1967.
26. Jolliffe CJ, Janssen I, 2006. Distribution of lipoproteins by age and gender in adolescents. Circulation. 114(10):1056-1062.
27. Obarzanek E et al., 2001. Long-term safety and efficacy of a cholesterol-lowering diet in children with elevated low-density lipoprotein cholesterol: seven-year results of the Dietary Intervention Study in Children (DISC). Pediatrics. 107(2):256-264.
28. Belay B, Belamarich PF, Tom-Revzon C, 2007. The use of statins in pediatrics: knowledge base, limitations, and future directions. Pediatrics. 119(2):370-380.
29. Edison RJ, Muenke M, 2004. Central nervous system and limb anomalies in case reports of first- trimester statin exposure. The New England Journal of Medicine. 350(15):1579-1582.
30. Lauer RM, Clarke WR, 1990. Use of cholesterol measurements in childhood for the prediction of adult hypercholesterolemia. The Muscatine Study. JAMA. 264(23):3034-3038.
31. Gidding SS et al., 2005. Dietary recommendations for children and adolescents: a guide for practitioners: consensus statement from the American Heart Association. Circulation. 112(13):2061- 2075.
32. Ray KK, Cannon CP, Ganz P, 2006. Beyond lipid lowering: What have we learned about the benefits of statins from the acute coronary syndromes trials? Am J Cardiol. 98(11A):18P-25P.
33. Fichtlscherer S et al., 2006. Differential effects of short-term lipid lowering with ezetimibe and statins on endothelial function in patients with CAD: clinical evidence for 'pleiotropic' functions of statin therapy. Eur.Heart J. 27(10):1182-1190.
34. Brotman DJ et al., 2005. In search of fewer independent risk factors. Arch Intern Med. 165(2):138- 145.
35. Ware WR, 2007. High cholesterol and coronary heart disease in younger men: The potential role of stress induced exaggerated blood pressure response. Med Hypotheses. e-published ahead of print.
36. Wald NJ, Law MR, 2003. A strategy to reduce cardiovascular disease by more than 80%. BMJ. 326(7404):1419.