The Cholesterol Wars: Steinberg Strikes Back
Academic Press, 2007
January 16, 2008
Review by Chris Masterjohn
Daniel Steinberg can be called nothing short of a true expert on cholesterol. He started his medical training in 1941 and spent over forty years researching atherosclerosis and producing hundreds of publications on the topic. He is one of the prime architects of the lipid hypothesis -- the idea that elevated levels of cholesterol and especially LDL in the blood cause heart disease.
In his new book, The Cholesterol Wars: The Skeptics vs. The Preponderance of the Evidence, Dr. Steinberg argues that cholesterol and lipoproteins have been "indicted, tried, and ultimately found guilty" of causing atherosclerosis and cardiovascular disease. He takes us through the entire history of this century-long trial and tells it from the perspective of one who was intimately involved in shaping it.
The majority of the science in the book is solid, and the arguments are generally well reasoned. So how is it that this conclusion can seem so far off?
In part, it is about how we use terms; in part, it is a matter of emphasis; in part, it is because Steinberg is actually right about some things that some cholesterol skeptics refuse to admit.
But, in part, there is also some bad science in the book. Although he is careful to emphasize the fact that he is supporting the lipid hypothesis rather than the diet-heart hypothesis -- that is, the idea that cholesterol in the blood, and not cholesterol and fats in the diet, is what causes heart disease -- he does make a limited argument in favor of the diet-heart idea, and it is one of the weaker points of the book. Most importantly, however, he actually argues that the statin trials are the final clincher showing that lowering cholesterol and lowering cholesterol alone reduces the risk of heart disease -- and, of course, statins by no means lower cholesterol alone. In fact, the statins constitute the greatest threat the lipid hypothesis has seen to date.
But first, let's go through the history of the lipid hypothesis as Steinberg conveys it.
In this Article
The History of the Lipid Hypothesis According to Steinberg
In the beginning of the twentieth century, researchers had tried to induce atherosclerosis in lab animals a number of different ways -- for example by injecting adrenalin, bacteria or their byproducts, or direct traumatization of blood vessels.
The medical field at the time considered atherosclerosis to be an inevitable consequence of aging. To test the hypothesis put forth by Nobel prize-winning microbiologist I. Metschinkow that dietary protein stimulated aging, the Russian researcher A. Ignatowski induced atherosclerosis in rabbits using diets of meat or milk and eggs. Other researchers showed eggs alone, with or without the white, and beef brains would also produce the disease. Finally Anitschkow and Chalatow produced the disease with pure cholesterol dissolved in sunflower oil, showing it was the cholesterol and not the protein at work.
Although the rabbits suffered from a variety of serious problems not usually encountered by humans such as cholesterol deposition in the internal organs, the atherosclerotic lesions were very similar to those seen in human atherosclerosis. They started as fatty streaks and eventually progressed to fibrous plaques with fibrous caps. The lipids accumulated in what appeared to be white blood cells infiltrating the vessel wall, as we now know occurs in humans. The distribution of lesions appeared to be determined by the direction and force of the flow of blood, as it is now believed to be in humans.
The extent of these lesions was proportional to the increase in blood cholesterol and the duration to which the rabbit was exposed to this increase. Injection of LDL and VLDL into the blood of the rabbits produced the same result, and lowering the blood cholesterol using the rabbit's normal herbivorous diet produced regression of the lipid accumulation but only partial regression of the connective tissue buildup in the more advanced lesions.
It's the Blood Level, Not the Diet
Later experiments verified the phenomenon in guinea pigs, but not in rats or dogs. It turned out that rats and dogs convert cholesterol to bile acids very efficiently and their blood cholesterol levels hardly change no matter how much cholesterol they consume in the diet. Blood cholesterol levels of rabbits, by contrast, shoot through the roof to 500-1,000 mg/dL when they consume cholesterol. If thyroid hormone is inhibited in dogs, however, dietary cholesterol produces a rise in blood cholesterol levels and subsequent atherosclerosis. In fact, Steinberg tells us, producing a sufficient increase in blood cholesterol levels has produced atherosclerosis in every species tested.
Correlations But Not Causation
Meanwhile, physicians were linking deposits of cholesterol in the skin called xanthomas with lesions in blood vessels and familial hypercholesterolemia. Later, Ancel Keys' Seven Countries Study (whose conclusions have been widely criticized in the cholesterol skeptic literature) and the Framingham Study showed correlations between cholesterol levels and heart disease but did not prove causation, and the Japanese Migrants Study showed that "some environmental factor" rather than genetics was the culprit.
Because in the 1950s and 1960s there was no way to lower cholesterol except by replacing saturated fats with polyunsaturated fats, researchers short-circuited the diet-blood and blood-heart connections to a diet-heart connection. Since responses of blood lipids to dietary lipids are so variable, Steinberg writes, this short-circuiting of the hypothesis led to a great deal of confusion.
Dietary trials at the time, often with various flaws, supported the lipid hypothesis but certainly not infallibly. In order to prove the hypothesis, researchers would have to lay down a basic science foundation showing the mechanism by which cholesterol causes heart disease and design powerful drugs to show unequivocally that lowering cholesterol reduces the risk of this disease.
The Basic Science of Pathogenesis
The basic science would show that HDL can remove cholesterol from the "foam cells" (lipid-loaded white blood cells) that enter the blood vessel wall during the formation of atherosclerotic lesions, while the remnants of triglyceride-rich VLDL and chylomicrons (another type of lipoprotein) can be taken up by white blood cells called macrophages in order to produce these same foam cells.
Most of the blame, however, would fall on LDL -- but LDL can only be taken up by macrophages once it is damaged by free radicals (oxidized) or free-floating sugars (glycated). LDL that is damaged in such a way -- but not LDL that has not been damaged -- is not only taken up by white blood cells but it attracts them to the blood vessel wall, immobilizes them, and initiates an inflammatory cascade. Because oxidized LDL can stimulate this inflammation, Steinberg argues, the dichotomy between lipids and inflammation is a false one.
Vitamin E and HDL protect LDL from being damaged.
Geneticists showed the cause of familial hypercholesterolemia to be a single mutation in the LDL receptor gene making the receptor non-functional. Since the cells could not take up LDL by this receptor, LDL would hang around in the blood and the level would rise. Steinberg argues that this finding proved that the elevated level of LDL was the cause of heart disease associated with familiar hypercholesterolemia, although he never entertains the possibility that the inability to bring that LDL into the cells is the cause -- for example, by increasing the amount of time an LDL molecule spends in circulation, and thereby making it more likely to oxidize.
Drugs, Drugs, and More Drugs
Early trials with cholesterol-lowering drugs in the 1960s were disappointing. Clofibrate reduced the risk of heart attacks but increased the level of liver, gall bladder, and intestinal diseases. Altogether, total mortality increased by twenty percent in one study. Another fibrate called gemfibrozil reduced the risk of heart attacks by 32 percent but had no effect on total mortality.
In the 1965 Coronary Drug Project, D-thyroxine (a synthetic version of thyroid hormone) produced arrhythmia. Both D-thyroxine and estrogen therapy increased the risk of heart disease in men. Nicotinic acid had no effect on total mortality during a five-year trial, although four years after the patients stopped the medication, the treated group had 11 percent fewer deaths. Even this drug, however, led to uncomfortable flushing symptoms and in some cases liver toxicity.
The Coronary Primary Prevention Trial
Things were not going so well. The "anti-cholesterol forces" needed a "you can't argue this" type of study -- a double-blind, randomized, placebo-controlled trial with a powerful but safe drug. In 1970, the National Institutes of Health (NIH) began laying the groundwork for the Coronary Primary Prevention Trial, a project it would spend 13 years and $150 million on. It established the Lipid Research Branch of the National Heart and Lung Institute and a national network of Lipid Research Clinics.
A double-blind diet study would have been impractical and forbiddingly expensive. Not to mention -- and Steinberg doesn't -- the two-year pilot study found no statistically significant difference in heart disease incidence and didn't bother reporting mortality.
The best drug available was a sandy powder called cholestyramine that would have to be taken in doses of two packets three times a day, mixed with water or juice. It caused bloating, constipation and diarrhea but it was free of systemic toxicity.
In 1973 the researchers at NIH set out to enroll around 3,800 men between the ages of 35 and 59 who were free of heart disease but had cholesterol levels above 265 mg/dL. After ten months, however, they had only recruited 74 patients. A change in strategy was in order. They hired a successful public relations firm that also handled PR for the arms manufacturer McDonnell Douglas. The firm allowed them to screen 10,000 of their employees. Lipid Research Clinics set up booths to offer free cholesterol screening at football games. By 1976, enrollment was complete.
At the end of the trial in 1984, 3,806 men had been followed for an average of 7.4 years. About 50 percent correctly identified which group they had been placed in, indicating that the double-blind design worked. The cholestyramine treatment reduced total cholesterol by 13.4 percent and LDL by 20.3 percent. It reduced coronary heart disease and nonfatal heart attacks by 19 percent, which was statistically significant by the p<0.05 level, but not by the more rigorous p<0.01 level they had initially hoped for (a fact that Steinberg doesn't discuss).
Although the result was not as impressive as they'd hoped for, the totality of the data was very compelling. The reduction in heart attacks was also met with a 20 percent reduction in anginal pain (p<0.01) and a 25 percent reduction in electrocardiogram (ECG) abnormalities (p<0.001).
Moreover, compliance had been measured throughout the study by the nurses who were distributing refills of the cholestyramine packets. Full compliance would have meant six packets per day, but the nurses gave out an average of only 4.2 packets. Among those who fully complied with the regimen, cholesterol levels were reduced 35 percent and heart attacks were reduced 49 percent. Thus, the totality of the data not only corroborated the heart attack figure with related endpoints like angina and ECG readings, but even showed a dose-response between the treatment dose, the cholesterol-lowering effect, and the heart disease outcome.
The seven percent reduction in total mortality, however, was not statistically significant. Although there was no statistically significant increase in any one endpoint, traumatic deaths such as accidents, suicide and homicide nearly tripled from four in the placebo group to eleven in the treatment group. Steinberg is quick to dismiss this finding because earlier clofibrate trials that increased total mortality did not increase traumatic deaths and because one of the deaths in the treatment group occurred because the patient was killed by a surprise burglar. This dismissal is another weak point in the book. Taking out the burglary, the traumatic deaths were still more than doubled. And Steinberg offers no discussion of the evidence connecting cholesterol-lowering and low cholesterol levels to violent suicide, depression and slowed reaction time that Anthony Colpo reviews in his book, The Great Cholesterol Con.
Still, the main point is whether heart disease was reduced, and the evidence Steinberg presents on this point is compelling.
The 1984 Consensus Conference
Steinberg presents the NIH Consensus Conference that took place that year as if it were very fair to dissenters. Mary Enig recalls it quite differently, noting that dissenters were allowed to speak (they gave three out of twenty presentations) but their views were not included in the pre-packaged consensus report. Steinberg does briefly note, however, that conference director Basil Rifkind sometimes had a "no-discussion-allowed" approach. (He was good at getting things done.)
The Consensus Conference endorsed the view that the lipid hypothesis had been proven. In doing so, it smoothed the way for the FDA to approve the coming statin drugs based on their cholesterol-lowering effect alone before they were tested against a real endpoint such as heart disease. According to Steinberg, however, it was these statins that really proved the lipid hypothesis and put the skepticism to rest.
Satins Prove the Lipid Hypothesis
The statin trials, Steinberg says, reduced cholesterol levels an average of 40 mg/dL and reduced major vascular events by 20 percent. The dose-response between cholesterol-lowering and heart disease reduction is slightly more with statins than with pre-statin drugs, but not much. The bulk of the benefit, Steinberg thus concludes, is due to their cholesterol-lowering effects. Statins are more effective at reducing heart disease simply because they are more effective at reducing cholesterol levels.
He even suggests, albeit with reservation, that if we could lower LDL to 57 mg/dL, we might be able to eliminate heart disease. He cites one finding that people with a genetic defect causing an overactive production of LDL receptors have low LDL from birth and an 88 percent reduced risk of heart disease. If we could start statin treatment even earlier than we currently do, then, we might prevent a lot more heart disease. The suggestion that we should put statins in the water supply is thus only half facetious and only half in gest -- although Steinberg says we should revisit this question only if a zero-side effect statin is invented and should focus on using exercise and diets low in total fat and saturated fat instead of pill-popping as we look toward the future.
Does the Conclusion Fit the Data?
Steinberg is careful to distinguish between the lipid hypothesis and the diet-heart hypothesis. Many writers, he says, lump both ideas under one banner, which leads to confusion. So far, so good -- evidence that dietary cholesterol does not cause atherosclerosis in rats or that saturated fat lowers the risk of heart disease in humans does not contradict the hypothesis that an elevated level of cholesterol in the blood causes atherosclerosis and heart disease.
But does the data Steinberg presents really indict an elevated level of cholesterol in the blood as the culprit? Chapter five on the basic mechanism of pathogenesis makes it clear that it is oxidized and glycated LDL that contribute to atherosclerosis. When LDL gets damaged, it is primarily the polyunsaturated fatty acids in the phospholipid membrane and the protein that weaves in between them that get damaged -- not the cholesterol that mostly lies deep in its core. In fact, the more cholesterol in the core, the larger and more buoyant the particle and the less likely the particle is to oxidize. That's why dietary cholesterol from eggs actually makes LDL safer.
Now, naturally a high level of LDL in the blood -- all things being equal -- will make that LDL more likely to oxidize. If the LDL level is high because more of it is being made, then more LDL particles means fewer antioxidants per particle. If the LDL level is high because of a genetic defect in the LDL receptor or because there is not enough thyroid hormone to make that LDL receptor work, then the LDL will spend more time in circulation and be exposed to sugars and free radicals for a longer period of time.
But all things are never equal. What if the level of LDL is high, but the level of antioxidants is also high? A 2005 study I reported on in my newsletter showed that the antioxidant resveratrol protected rabbits against the negative effects of massive cholesterol feeding without having any effect on the total cholesterol level.
So why doesn't Steinberg conclude that it is high blood sugar, free radicals, deficient antioxidants, subclinical hypothyroidism, or excess polyunsaturated fatty acids that have been "indicted, tried, and ultimately found guilty" of causing heart disease?
Certainly it is not cholesterol who is the primary culprit, when cholesterol sits rather innocently or perhaps protectively, mostly at the inner core of the LDL particle, while the sugars and proteins and polyunsaturated fats are causing trouble on the outer surface.
Statins Do Not Prove the Lipid Hypothesis — They Do the Opposite!
The strangest claim in the book is that statins prove the lipid hypothesis. In fact, it is the statins that have shown us the critical importance of Rho activation to heart disease and called into question decades of research associating this disease with cholesterol.
The main enzyme that controls the production of cholesterol is HMG CoA reductase. As you can see from this chart, HMG CoA reductase produces a compound called mevalonate. The cell can use mevalonate to make cholesterol but it can also use it to activate an enzyme called Rho.
Another enzyme called squalene synthase regulates the balance between these two uses of mevalonate: stimulating squalene synthase diverts mevalonate into cholesterol production while inhibiting squalene synthase diverts mevalonate into Rho activation.
Rho induces a stress response in the cell, causing a reorganization of its basic structure and strongly inhibiting nitric oxide synthase, the enzyme that produces nitric oxide.
Nitric oxide is a gas that protects against heart disease at every level -- it increases blood flow and vessel dilation, decreases the adhesion of white blood cells to the vessel wall, inhibits the migration of smooth muscle cells to the site of an atherosclerotic lesion, and decreases the formation of blood clots. So when Rho gets turned on, nitric oxide gets inhibited, and atherosclerosis begins. It has nothing to do with cholesterol.
Yet it certainly correlates with cholesterol. Inflammation, for example, stimulates HMG CoA reductase but inhibits squalene synthase. This causes a moderate increase in cholesterol levels and a large increase in Rho activation. Thus, you'd expect to find an indirect correlation between cholesterol levels and heart disease in the population because they are both correlated with inflammation.
Statins inhibit HMG CoA reductase. In doing so, they reduce cholesterol levels and Rho activation at the same time -- since they reduce both of these by the same exact mechanism, the degree to which they reduce one will correlate with the degree to which they reduce the other.
And what about cholestyramine? Cholestyramine results in a massive increase in squalene synthase activity. It lowers cholesterol by binding up bile acids and causing the body to use up its store of cholesterol to make more of them. But it also causes the body to use up its store of mevalonate to make more cholesterol! The result is a lower level of cholesterol and -- although no one has directly tested it -- almost certainly a lower level of Rho activation.
Steinberg argues that because the relationship between cholesterol lowering and heart disease reduction is only slightly higher with statins than it was with the pre-statin drugs, most of the heart disease reduction on statins is due to cholesterol lowering. This assumes that all the previous drugs only lowered cholesterol. But they too almost certainly lowered Rho activation as well.
Ironically, what statins have actually taught us is that Rho activation and nitric oxide synthase inhibition have probably been confounding decades of research on cholesterol -- from epidemiological studies to pre-statin drug trials to the statin trials themselves, confounding even the analyses attempting to differentiate the cholesterol-lowering effect of statins from their "pleiotropic" effects. Rather than proving the lipid hypothesis, statins cast doubt on a great deal of the support it had previously gained.
Are Lipoproteins Innocent?
That said, are lipoproteins vindicated? Not completely. We know from basic molecular biology that oxidized LDL itself inhibits nitric oxide production and activity. We know it initiates inflammation. We know it loads itself into foam cells and we know it doesn't play any "protective" role in arterial plaque -- instead, it contributes to the buildup of connective tissue matrix and the weakening of fibrous caps.
But we also know that nitric oxide protects LDL from oxidizing. So do HDL, vitamin E, polyphenols, and other important antioxidants.
So which makes the greater contribution? Rho activation or oxidized and glycated LDL? That is what we do not know.
Diet — Where the Book Really Goes Wrong
Although Dr. Steinberg emphasizes that he is dealing with the lipid hypothesis rather than the diet-heart hypothesis, he nevertheless makes a substantial and unfortunate detour in chapter three to defend the American Heart Association for its recommendations to reduce the intake total and saturated fat. He further concludes the book by looking toward a future where we use exercise and diet instead of pills to keep people healthy. While I would wholeheartedly agree with this, he again identifies saturated fat as the dietary villain. Quite the opposite is true.
Steinberg supports his stance against saturated fat with a number of controlled metabolic ward studies showing that substitution of saturated fat for polyunsaturated fat in liquid milk shake formulas increases cholesterol levels, three higher-quality substitution studies that use actual disease endpoints, and four lower-quality studies using actual disease endpoints that he ranks as flawed or even "seriously flawed" but that he believes nevertheless add to the support for the diet-heart hypothesis.
The studies using actual disease endpoints are the ones we care about. The problem is that the evidence is not nearly as good as Steinberg says it is, and he leaves out a number of important trials.
Any reader of The Cholesterol Wars should cross-reference chapter three with chapter eight of Anthony Colpo's The Great Cholesterol Con. Colpo shows definitively that the totality of controlled experimental studies fails to indict saturated fats and casts serious doubt on the healthfulness of polyunsaturated vegetable oils.
The Three Studies
The three higher-quality studies that Steinberg presents are the Paul Leren Oslo Diet-Heart Study, the Wadsworth Veterans Administration Hospital Study, and the Finnish Mental Hospitals Study.
The Oslo Study replaced saturated fat with polyunsaturated fat, including a pint of soybean oil per week. It lasted five years and produced no difference in all-cause mortality but decreased the incidence of second heart attacks. But the control group started out with a higher number of older and overweight subjects; although it started with the same proportion of heavy smokers as the treatment group, it ended with twice as many heavy smokers. Members of the treatment group were counseled to eliminated their intake of margarine; increase their intake of fruits, vegetables, and fish; and were provided with free sardines canned in cod liver oil. There are far too many confounding variables to suggest this study actually indicts saturated fat.
The Wadsworth Veterans Administration Hospital Study compared meals made with saturated animal fats to meals made with polyunsaturated vegetable oils among institutionalized veterans for eight years. There was a reduction in cardiovascular events, but an increase in cancer of the same magnitude.
Colpo points out that the autopsies showed little difference in atherosclerosis between the two groups except somewhat more aortal plaque in the group that ate the vegetable oil. He suggests that the difference in the rate of cardiovascular events was due to the higher rate of heavy smoking in the control group. (Smoking increases the risk of spasms that can lead to heart attacks, independent of atherosclerosis.) The proper interpretation of this study, then, is that members of the treatment group had a lower rate of cardiovascular events because they smoked less, but had more atherosclerosis and cancer despite smoking less -- probably because of the vegetable oil they were eating!
The Finnish Mental Hospitals Study used "filled milk" -- milk whose fat was replaced with soybean oil -- and a polyunsaturated margarine in place of butter. In one hospital, the patients ate their regular diet for six years and then the treatment diet for six years; in the other, they ate the treatment diet first and the regular diet afterwards. In both hospitals total and cardiac mortality was lower on the polyunsaturate-rich diet, but the effect was much more pronounced in the hospital that used this diet first, perhaps because of the confounding effect of age.
This study is certainly more compelling than the other two, but it has one massive flaw: the subjects came and went; if they left for good, they were excluded from the data; if they left and came back, they were included; even if they were newly enrolled just a month before the data was collected, they were still included! A study with this type of irresponsible and possibly even malicious design certainly cannot stand on its own as evidence indicting saturated fat.
Yet More Studies
Steinberg cites four other studies to which he attributes a lower level of quality, each of which are unconvincing at best: the Lester Morrison Study, the Anti-Coronary Club Study, the Bierenbaum St. Vincent Hospital Study, and the British Medical Research Council Study.
The Lester Morrison Study reduced total mortality with a low-fat, low-calorie diet that was high in protein and supplemented with wheat germ and brewer's yeast, supplying plenty of B vitamins, vitamin E and selenium. Low-fat means low in the polyunsaturated fats that the body cannot make and high in the saturated fats the body makes itself -- and with plenty of extra vitamins and minerals to protect against mortality.
The Anti-Coronary Club Study replaced animal fat with vegetable oil. It was widely hailed for its reduction in non-fatal heart disease events that included soft endpoints like angina and ECG abnormalities. But the fact that the treatment more than doubled total mortality and increased heart disease mortality from zero to one percent (zero to nine deaths) was widely ignored.
The Birenbaum St. Vincent Hospital Study found no difference between a diet that was half corn and half safflower oils and a diet that was half peanut and half coconut oils except a 25 percent higher total mortality rate on the first, more unsaturated diet. They pooled the data together and compared it to another group with higher cholesterol levels and, looking backward, determined that the group with higher cholesterol had more heart disease over the previous five years. Steinberg calls this second analysis a "limited study with serious flaws" that nevertheless supports the lipid hypothesis, but it obviously has nothing to do with saturated fat.
The British Medical Research Council Study generated no statistically significant findings. Patients were counseled to reduce saturated fat intake and consume three ounces of soybean oil per day. Cardiac events including angina were slightly lower in the treatment group but total and cardiac mortality were the same. Not terribly impressive.
Leaving a Few Out...
Whereas Steinberg cites only seven studies in his treatment of this issue, Colpo cites nineteen. Among the research Steinberg leaves out stand the following two embarrassing studies:
- Rose, et al. (1965): Replacing animal fat with corn oil for two years lowered serum cholesterol by 23 mg/dL but quadrupled cardiac and total mortality.
- Sydney Diet-Heart Study (1978): Replacing animal fat with vegetable fat for five years lowered cholesterol by five percent but increased total mortality by 50 percent.
Granted, Steinberg only means to cover pre-1970s studies -- but missing the extremely embarrassing Rose study is hardly an excusable oversight.
In his own analysis, Colpo cites a number of other unsuccessful trials attempting to reduce heart disease by reducing the intake of total or saturated fat or replacing saturated fat with vegetable oil, sometimes despite a large decrease in serum cholesterol. For example, Ball et al. (1965) were able to reduce serum cholesterol by 25 mg/dL with a diet low in total and saturated fat, but the treatment had no effect on the risk of heart disease.
After reviewing all the evidence, Colpo concludes:
If saturated fats caused even a portion of the damage for which they are frequently blamed, their negative effects should be readily and repeatedly demonstrable in controlled clinical trials. However, after excluding the results of the poorly designed and sloppily conducted northern European studies, it quickly becomes apparent that there does not exist a single tightly controlled trial which shows that saturated fat restriction can save even a single life.
Some of these trials, in fact, suggest the exact opposite. The Rose et al., Anti-Coronary Club and Sydney Diet Heart studies all showed significant increases in overall mortality from replacing animal fats with omega-6-rich vegetable fats. The longest-running study focusing on saturated fat restriction (the Los Angeles Veterans study) showed a significant increase in cancer mortality among the intervention subjects -- despite their lower rate of smoking! . . .
It must again be emphasized that the dietary intervention studies discussed in this chapter represent sixty years' worth of intensive research, the expenditure of hundreds of millions of dollars in public funds, and an enormous amount of time and effort. The reason this massive undertaking has failed miserably to find any causative role for highly saturated animal or tropical fats in the development of CHD should by now be obvious -- there is none!
Low-Fat Diets Are High in Saturated Fat
Steinberg also suggests that reductions in total fat are similar in effect to reductions in saturated fat. The reasoning seems simple enough -- if you reduce your fat intake, some of that fat is saturated, so you will necessarily be eating less saturated fat.
But the exact opposite is true. The body cannot make polyunsaturated fats but readily makes saturated and monounsaturated fats from carbohydrate. A low-fat diet is low in polyunsaturated fats and effectively high in the saturated fats that the body will make itself.
Consider George Burr's (discoverer of the essential fatty acids) description of the first attempt to induce essential fatty acid deficiency in an adult by eating a diet extremely low in fat:
A much more sophisticated experiment was done when biochemist W. R. Brown volunteered to live for six months on a diet extremely low in fat. He was clinically well throughout the entire period, not having even a common cold. There was a marked absence of fatigue. Attacks of migraine subsided completely. The respiratory quotient rose markedly after a meal. Blood total lipids increased but unsaturation decreased 25%. Linoleic and arachidonic acids decreased about 50%.1
Since saturated fatty acids are not vulnerable to oxidative damage and since it is the unsaturated fatty acids in the LDL membrane that oxidize, we should expect a diet rich in saturated fat and low in polyunsaturated fat to protect LDL from damage. Any benefits from low-fat diets should be seen as benefits of polyunsaturated fat restriction, not saturated fat restriction.
Reducing the intake of polyunsaturated fat also sensitizes the body to thyroid hormone.2 Thyroid hormone helps take up LDL into cells through the LDL receptor, so a diet rich in saturated fats and low in polyunsaturated fats may also decrease the amount of time LDL spends circulating in the blood and thereby decrease the time it spends exposed to free radicals and sugars and its likelihood of oxidizing and causing damage.
The conclusions we reach about diet show why it is not simply semantics but critically meaningful to distinguish between LDL as the culprit and oxidized LDL as one of the culprits.
Tightly controlled metabolic ward studies did in fact show that substituting polyunsaturated fats for saturated fats reduces LDL and total cholesterol. If LDL or total cholesterol were the primary villains, it would follow directly from this that making this substitution would also reduce the risk of heart disease. But if oxidized LDL was merely one of several villains, one would have to ask tougher questions: What effect does this substitution have on oxidation? On inflammation? On glycation?
Steinberg never asks these questions. He recognizes that it is oxidized and glycated LDL we need to worry about, but somehow he considers this a confirmation of the hypothesis that total and LDL cholesterol are the villains rather than a refutation or at least a modification of this hypothesis. So while he acknowledges this fact in chapter five, he otherwise ignores it.
When we ask those critical questions, however, we come to very different conclusions about diet. We should be minimizing polyunsaturated fats, not saturated fats; we should be maximizing antioxidants and anti-glycating agents from meat that is not overcooked, grass-fed animal fats, CoQ10-rich heart and vitamin-rich liver, vitamin E-rich palm oil, polyphenol-rich virgin coconut oil, fresh fruits, vegetables and nuts, and freshly ground grains; and we should eat plenty of eggs, which increase the size and safety of LDL particles and load them up with protective carotenoids.
And yes, Steinberg is certainly right on one thing -- we should get plenty of exercise.
Conclusions About Cholesterol and Who Should Read This Book
All criticism aside for a moment, this book is an important addition to the cholesterol debate. Steinberg presents a compelling case for the relevancy of animal studies and the success of the Coronary Primary Prevention Trial. His two chapters on the basic science of cholesterol and lipoproteins add a great deal to the debate, which is too often dominated by epidemiology.
Steinberg presents the history of the controversy as one who was intimately involved with it. It is fascinating to read how intertwined the development of the lipid hypothesis has been with the development of modern biological science itself. For example, the LDL receptor was the first of the cell surface receptors discovered and the meta-analysis of cholesterol-lowering drug trials at the 1984 Consensus Conference was one of the first if not the first use of the meta-analysis as a statistical tool.
It is also interesting to contemplate the sheer movement of cash (representing real economic resources) into these studies and to consider how fruitless the use of these resources has actually been. The technology developed by private industry has saved more lives by making heart disease less fatal. Meanwhile, cattle-herding tribes like the Masai have protected themselves against heart disease far more effectively than we have, having neither hundreds of millions of dollars for even a single placebo-controlled trial nor any of the fancy technology that industrial capitalism produces.
Steinberg aims the book primarily at the medical community, but anyone who is reading books like Ravnskov's The Cholesterol Myths, Colpo's The Great Cholesterol Con, or Kendrick's book by the same title should read this book to get the other side of the story.
Conversely, no one should read this book without also reading at least one of the above selections from the leading skeptics. Steinberg often includes important details from studies that others leave out, but he also often leaves out important details himself, or even whole studies that disagree with his conclusions.
Ultimately, Steinberg fails to produce a convincing argument that total and LDL cholesterol are the primary villains in heart disease, but he does make a convincing case that they are not completely irrelevant. The basic science showing that oxidized and glycated LDL can not only accumulate into foam cells but initiate and aggravate the inflammatory cascade shows clearly that there is nothing protective about this lipid accumulation, and that there is no analogy between it and firemen being found at the scene of a fire.
At the same time, one would get the idea from reading The Cholesterol Wars that there is actually good science showing that we should reduce our intake of butter and use vegetable oil instead! And nothing could be further from the truth.
As always, it is best to read both sides. Thanks to Dr. Steinberg, we now can.
1. Holman RT. George O. Burr and the Discovery of Essential Fatty Acids. J Nutr. 1988;118:535-540.
2. Panos TC. Effects of fat deficiency on the endocrine system in rats. J Pediat. 1963;63:1197.