The Lipid Hypothesis – Closing in on the Truth

Estimated reading time: 6 minutes

Today was a big day at the annual conference of the American Heart Association in Chicago. The results of the long and eagerly awaited IMPROVE-IT trial were presented.

You may think IMPROVE-IT is just another ordinary study of the efficacy of a particular cholesterol lowering drug. But it’s more than that because it tests the important and heavily debated lipid hypothesis.

The Lipid Hypothesis - Closing in on the Truth

The lipid hypothesis implies that cholesterol, particularly LDL-cholesterol plays a key role in causing atherosclerosis and coronary heart disease. It means that when it comes to heart disease, measures that elevate blood levels of LDL-cholesterol are usually bad and measures that lower it are good.

Initially, the lipid hypothesis was mainly supported by observational data showing a linear relationship between blood cholesterol and mortality from coronary heart disease (1).

The association between cholesterol and overall mortality appears more complicated. Studies have shown a correlation between cholesterol levels and overall death rate in young and middle-aged people but not among the elderly (2,3). However, the mortality curve appears J-shaped which means that those with the lowest cholesterol levels have increased mortality.

It has also been suggested that the relationship between cholesterol and cardiovascular risk may not be the same for men and women. A Norwegian study (4) found an inverse relationship between cholesterol levels and mortality among women, for whom, according to the authors, moderately raised cholesterol may prove to be not only harmless but even beneficial.


Intervention Trials

The main limitation of observational studies is that the variable being tested is not randomly distributed. In other words, there may be underlying reasons for high and low cholesterol. In theory, these reasons could affect outcome, and therefore cholesterol might be nothing more than a surrogate.

In controlled intervention trials, a hypothesis is tested using specific methods designed to examine the efficacy of an intervention. Drug trials where half of the patients are treated with a certain drug and half are treated with placebo are a typical example of controlled intervention trials.

In fact, the validity of the lipid hypothesis has been tested in controlled interventional trials.

Lifestyle and Cholesterol

The largest controlled intervention trial on diet and heart disease to date, the Women’s Health Initiative (4) randomly assigned more than 48 thousand women, 50 – 79 years old, to a low-fat intervention or a comparison group. LDL-C was significantly lowered in the intervention group compared to the comparison group. Nonetheless, after six years of follow-up, there were no differences between the groups in the incidence of coronary heart disease and stroke.

The MRFIT trial (5) evaluated 12,866 high-risk middle-aged men who were randomly assigned either to a special intervention program consisting of stepped-care treatment for high blood pressure, counseling for cigarette smoking, and dietary advice for lowering blood cholesterol levels or to their usual sources of health care in the community. LDL-C was significantly lowered in the special intervention group compared to the “usual care” group. However, during a follow-up of seven years, there was no significant difference in total death rates between the groups and no differences in the number of deaths from heart disease.

The results of these two large trials strongly indicate that lifestyle measures aimed at lowering LDL-C do not improve survival or reduce mortality from coronary heart disease.


12165790_mStatins and Cholesterol

Most experts agree that the use of statins has revolutionized the treatment of coronary heart disease. Double-blind placebo-controlled trials have shown that statins reduce mortality and lower the risk of future cardiovascular events.

Statins effectively lower blood cholesterol, particularly LDL cholesterol.

If the lipid hypothesis is correct, most of the efficacy of statins is due to their cholesterol-lowering effects.

But statins might also work through other mechanisms.

The Jupiter trial (6) suggested that treatment with statins may have beneficial effects on people with relatively low levels of LDL cholesterol. The individuals who participated in this trial all had elevated levels of hs-CRP which is a marker of inflammation.

These results may suggest that cholesterol is only a byproduct and that the efficacy of statins may be explained by other mechanisms, such as reducing inflammation.

Statins are potent inhibitors of cholesterol biosynthesis. However, the overall benefits observed with statins appear to be greater than what might be expected from changes in lipid levels alone, suggesting effects beyond cholesterol lowering.

In fact, recent studies (7) indicate that some of the cholesterol-independent or “pleiotropic” effects of statins involve improving endothelial function, enhancing the stability of atherosclerotic plaques, decreasing oxidative stress and inflammation, and inhibiting blood clotting mechanisms.


Ezetimibe is a drug that lowers cholesterol by a mechanism that is very different from that of statins. Ezetimibe blocks the absorption of cholesterol from the small intestine. This leads to an increased availability of LDL-receptors on the surface of cells leading to increased uptake of cholesterol by liver cells and therefore lowering of blood cholesterol levels.

When added to statin therapy ezetimibe generally lowers cholesterol by an additional 20-25% over the statin alone.

In the US ezetimibe is marketed by Merck as Zetia (ezetimibe alone) and Vytorin (ezetimibe in combination with a statin called simvastatin).


The IMPROVE-IT study tested the hypothesis that further lowering of LDL cholesterol by adding ezetimibe to statin therapy in patients with coronary artery disease will improve outcome.

IMPROVE-IT began enrolling patients with acute coronary syndrome in 2005. It was designed to evaluate the effects of concomitant simvastatin and ezetimibe therapy compared to simvastatin therapy alone on the composite endpoint of cardiovascular death, nonfatal myocardial infarction, rehospitalization for unstable angina, coronary revascularization, or stroke.

A total of 18,144 high-risk patients were randomized within ten days of an acute coronary event to either ezetimibe or placebo on top of a statin. The patients had LDL cholesterol levels between 50-125 mg/dL (1.3-3.2 mmol/L), or between 50-100 mg/dL 1.3-2.6 mmol/L) if already on a cholesterol lowering drug. The average LDL cholesterol at baseline was 95 mg/dL (2.5 mmol/L).

The average age of the patients was 64, and about a fourth were women. Patients were followed for an average of six years. There were 5,250 primary endpoint events (cardiovascular death, myocardial infarction, hospital admission for unstable angina, coronary revascularization more than a month after randomization, or stroke).

Primary endpoint events occurred in 34.7% of the control group versus 32.7% of the treatment group. This represents a 6.4% relative risk reduction and a 2% absolute risk reduction. This means that 50 patients would need to be treated for seven years to prevent one event. Interestingly 49 out of 50 patients treated saw no benefit.

There was no difference between the groups in overall death rate or death rate from cardiovascular disease, but there were significant reductions in the rate of myocardial infarction (13%, p =0.002), stroke (14%, p=0.052), and ischemic stroke (21%, p=0.008). The effect was similar across subgroups, except for diabetics, who had a larger benefit than those without diabetes.

As expected, LDL cholesterol was significantly lowered in the treatment arm; median levels were 69.9 mg/dL (1.8 mmol/L) in the control group versus 53.2 mg/dL (1.4 mmol/L) in the treatment group. No safety issues were found, and there were no differences between the groups in cancer, muscle or gall-bladder related events.

Thus, although the effect was modest, the results of the IMPROVE-IT trial suggest that lowering cholesterol more than can be achieved with statins may improve clinical outcome.

Interestingly, however, in this high-risk group, the lowering of LDL cholesterol did not significantly affect mortality. Furthermore, 98 percent of the patients didn’t benefit from treatment.

Anyhow, there is no denying that the lipid hypothesis is still alive and well.


66 thoughts on “The Lipid Hypothesis – Closing in on the Truth”

  1. Nice summary Axel. I am bothered by the fact that 42% of subjects dropped out of the trial. I wonder if the results would have been larger (or smaller) were that not the case.

    • I agree Barbara. It’s a large percentage. I think a secondary analysis, addressing the results of only the patients who stayed on the study drug, is scheduled for presentation at the AHA meeting tomorrow.

    • It is surprisingly large, indeed. However, it seems there was no significant difference between the control and the treatment arm in this sense?

      Axel, anything on the per protocol analysis?

  2. Wondering if I should take the Lipitor my doc wants me to take. Female, age 49. 5’5 128lbs., nonsmoker, nondrinker, exercise (cardio and strength) 5-6 days a week. LDL-P 2,283. LDL-C 126, hDL 56. No other risk factors other than genetics. Will be able to send all numbers when I return home to view. My dad takes a statin.

    • Gina, I hope you were not “talked into” taking a statin. There is NO reason for you to alter your LDL and HDL through drug use. Your cholesterol markers are bad due to your diet; Too many carbs and not enough fat. Grains are especially problematic. Grains turn to sugar, and your numbers show you are eating too many. Statins are no the answer and SHAME ON YOUR DOC for suggesting them. The side effects of statins are awful, not to mention they have been known to cause heart disease. Please go to FB and check out a few pages by reputable docs: Dr. William Davis, Dr. David Perlmutter, Dr. Joseph Mercola, Dr. Stephen Sinatra, and Dr. Eric Westman, who wrote “Cholesterol Clarity”. Educate yourself, because you can’t count on your typical “M.D.” to do more than write an Rx. It’s sad.

  3. Axel,

    thanks for the article, I like particularly the last line 🙂 But, I am cannot help being puzzled. What do you mean when you write that LDL was significantly lowered in the WHI study? Have either you or I missed something? The study authors write the following:

    “Because there are no apparent changes that would have mitigated a potentially favorable effect on CVD, the lack of an appreciable CVD effect maybe attributable to the limited decrease (only 2.7 mg/dL [0.07 mmol/L]) in LDL-C level, as well as the modest differences in other potentially favorable dietary components. Based on a large body of evidence from LDL-C–lowering trials, this magnitude of change in LDL-C level would be predicted to produce only a small (2%-4%) decrease in CVD risk, a value far below the power for detection in the current study. As delivered, the dietary intervention was not expected to have substantial effects on lipoprotein levels, but it is possible that a diet specifically lower in saturated and trans fat combined with increased intakes of vegetables ,fruits, and grains might have led to a decrease in CVD risk.”,

    “Compared with those in the entire comparison group, a trend was observed toward reduction of CHD risk among those in the intervention group who reached the lowest levels of saturated fat (HR, 0.81; 95% CI, 0.69-0.96 in the group that consumed <6.1% energy; P<.001 [adjusted HR, 0.82; 95%CI, 0.67-0.99; P=.05])… While these additional analyses are subject to residual confounding because of reporting bias or the lack of a comparable comparison group, some confidence in their validity is supported by parallel patterns of LDL-C reductions in participants stratified by changes in saturated fat at year 1 (−10.1; 95% CI, −13.5 to −6.6 mg/dL[0.26; 95% CI, −0.36 to −0.17 mmol/L]in the quartile with the greatest reduction;P=.005 for trend)”,

    In the Women’s Health Initiative the control group who were advised to lower fat intake significantly increased the intake of sugar and showed trends towards lower body weight and waist circumference. If anything there was a slight trend towards a lower risk of diabetes in the control group.

    • Yes, In the WHI trial LDL was significantly lowered in the treatment arm, just take a look at the abstract: Low-density lipoprotein cholesterol levels, diastolic blood pressure, and factor VIIc levels were significantly reduced by 3.55 mg/dL, 0.31 mm Hg, and 4.29%, respectively;

      You may argue that the effect on LDL cholesterol was small. That’s fine and I agree.

      But you asked what I meant. I only meant it was statistically significant. And it was.

      • Ok Doc,

        thanks for the explanation. It just that the LDL hypothesis is not about statistically lower LDL as such, it’s about lowering LDL meaningfully. This is at least how I understand it.

        I have 3 more points:

        1) Almost every health parameter show more or less J or U-shaped-curve: BMI, blood-pressure, alcohol intake, possible even glucose levels. I think this is an important issue to point out. Old people with cancer and other chronic disease often looses radically weight at the last years. While predisposing to death they have low blood-pressure and very low BMI. However, we should be able to differentiate between snap-shots of time and life-time exposure.

        2) A skeptic to the “beyond LDL properties” of statins might claim all the beneficial properties are due to LDL lowering itself. A recent study showed that mechanically removing LDL cholesterol from the blood reduces arterial inflammation in FH patients

        3) Genetic evidence ties LDL exceptionally strongly to CHD and mortality risk. In addition to large meta-analysis of mendelian randomization studies, a recent study found that mutations that disrupt the Ezetemibe targeted NPC1L1 function is associated w/ reduced LDL cholesterol levels and a reduced CVD risk. A 12mg/dl decline in life-time LDL levels compared to Western norm, translated into 53% reduction in cardiadic events.

  4. I readily grant that I may be missing something here, but if

    1. There was no difference between the groups in overall death rate, or death rate from cardiovascular disease,

    2. There were significant reductions in the rate of myocardial infarction, stroke, and ischemic stroke in the treatment group, and

    3.No safety issues were found and there were no differences between the groups in cancer,

    then what cause of death increased in the treatment group, so that that group died at the same rate as the control group?

    • Mark
      I assume the reductions in MI and stroke were reductions in non-fatal events.
      You’re right, overall mortality and cardiovascular mortality were similar in both groups.
      We therefore have to assume that mortality from non-cardiovascular causes was similar as well.

  5. The difference in LDL-c in MRFIT was also very small between the groups, and the mortality rate was so low that it was greatly underpowered. Besides, some of the participants were already on cholesterol-lowering diets, unlike e.g. the Oslo study.

    Anyway, I don’t think the findins on Ezetimibe will change much in practise, it is already routinely used. But it supports the lipid hypothesis, and the results from PSCK9-inhibitor studies may bolster it further.

    • I wonder if the small effect on LDL cholesterol seen in the WHI and MRFIT reflects what can be achieved by lifestyle intervention in terms of lowering LDL cholesterol. In other words; targeting LDL cholesterol through diet will not affect LDL cholesterol enough to be clinically meaningful (unless everybody goes “Ornish”).

      Obviously dietary choices will never get our LDL cholesterol levels as low as statins and PCSK9-inhibitors can. So, when it comes to lowering of LDL cholesterol, lifestyle measures don’t stand a chance.

      So, the survival of the lipid hypothesis will continue to be the lifeblood of the pharmaceutical industry for years to come.

      • Are there any studies underway on PCSK-9 alone? I have only seen those in combination with statins. Seems like that would put the controversy to rest once and for all.

      • Elliot
        All the ongoing trials on people with known cardiovascular disease are with PCSK9-inhibitors added to statin therapy. It’s not considered ethical to not give statins in this population.
        There are probably some ongoing studies on PCSK9 inhibitors alone in other populations. For example it has been suggested that PCSK9 inhibitors may be useful in those who are statin intolerant.

      • Doc,

        “In other words; targeting LDL cholesterol through diet will not affect LDL cholesterol enough to be clinically meaningful (unless everybody goes “Ornish”)”.

        This seems plausible. However, it’s about the cumulative burden. Modest changes in diet will yield modest changes in lipid profile. The intake of SFA at the highest quartile of consumption in rural Japan of the 1970s was 3 grams less per day compared to the lowest quartile of consumption in Nurse’s health study. The mean age at the randomization in statin trials is 63 and the mean duration few years. I think the same goes for diet trials. Within this narrow window at later stages of life, a large reduction in LDL is needed for meaningful results. Given a population with life-long exposure to high SFA diet, a much broader time period is needed. However, if we start early, even modest difference in LDL should prove to be beneficial. This is the only conclusion one can draw from the genetic data.

      • I do not agree with your comment: “targeting LDL cholesterol through diet will not affect LDL cholesterol enough to be clinically meaningful (unless everybody goes “Ornish”).”

        On the contrary, LCHF can have a remarkable impact on LDL in patients with metabolic syndrome.

        Total chol. 7.12 3.89
        HDL 1.08 1.25
        TG 2.76 1.15
        LDL 4.8 2.1

        A familiy physician in Iceland with a history of CVD and metabolic syndrome published an article in one of the newspapers where he described his experience with LCHF. The table above shows his labresults before and after. The column on the left is from the 10th of June, just when he started. The column on the right is 2 weeks later!
        He has done several measurements since then that have shown that his lipid profile had stabilized at this level.

        He is far from being the only example of this, in my practice I have seen quite a few other patients getting similar results.

        This article was published in JAMA 1967 that describes this carbohydrate-sensitive hyperglyceridemia:

      • You’re right Gudmundur. Thanks.

        I realize I didn’t make myself clear enough. Of course I too have seen people influence their LDL cholesterol levels through diet. I was merely referring to the clinical trials where it has turned out be difficult to get a meaningful lowering of LDL cholesterol through lifestyle or dietary measures, at least if you compare it to the effect of statins

        On the other hand we should keep in mind that risk can be influenced without lowering LDL-cholesterol. Remember the LYON-Heart study were men with coronary artery disease who were randomized to a Mediterranean diet fared better than those randomized to the prudent low fat diet. However, LDL-cholesterol stayed the same in both groups.

        So, success isn’t always measured by LDL cholesterol.

      • Regarding the Lyon study – according to the 1999 paper, high cholesterol and blood pressure were indeed significant predictors of events. “… each increase of 1 mmol/L of total cholesterol increased the risk of recurrence by 20% to 30%”.

        So, “neither the Mediterranean dietary pattern nor any major bias has altered the usual and expected relationships between the major risk factors of CHD and recurrence.”

  6. The low LDL levels in these patients at baseline seem to suggest that coronary cases already have lower LDL than the healthy population of the same age.

      • Err, you do realize that the article you linked puts the focus on LDL & other classical risk factors, stating that age as a risk factor needs to be considered differently than now, from the point of view of cumulative risk due to life-long high LDL levels?

        “On the contrary, we will try to demonstrate that the reason that the causes of vascular disease are weak risk factors for clinical events is primarily a function of how we treat age as a determinant of vascular disease”

        And “modest”, please.

      • Arghh, goddamn I suck at reading! I failed to notice the word “not” in this

        “but that doesn’t mean it’s not a necessary cause of atherosclerosis”

        part of your message. Scratch my message above, will you. Please. 🙂

      • I think the word “cause” and the notion of causality is fooling us here.
        If you drive drunk and crash your car, you are more likely to die going 100k than going 90k.
        But the primary cause of your death is surely the decision to drive drunk, and the immediate cause is losing control (probably because you’re drunk).
        This sort of confusion is why epidemiologists use the criteria defined by Bradford Hill back in the 60s.
        Applying these criteria to diet isn’t done often enough, but this is an example.

    • George
      IMPROVE-IT selected patients with LDL cholesterol below 125 mg/dl (3.2 mmol/L) or below 100 (2.6 mmol/L), i.e. patients with higher LDL cholesterol levels were excluded.
      These were the two main inclusion criteria:
      1. Clinically stable subjects may be eligible to enroll within 10 days following hospital admission with high-risk acute coronary syndrome (either STEMI or Non-STEMI or unstable angina)
      2. Subjects not taking a statin must have an LDL-C of 125 mg/dl or less. Subjects taking a statin must have an LDL-C of 100 mg/dl or less.

      • Thanks!

        If CHD hits people with already low LDL, then either it’s specific to a type of LDL (pattern B, small dense or whatever) and/or LDL is not causal, but is taken hostage by the disease process at any concentration, with slight reductions in its involvement as levels drop.

        Do we know anything about changes in other risk factors? TG/HDL ratio? ApoA/B?

        “Ezetimibe also significantly improved levels of plasma total cholesterol, apolipoprotein B, high-density lipoprotein(2)-cholesterol and lipoprotein(a), and elicited a trend toward lower triglyceride levels.”

        If that’s so, the jury is still out on LDL level per se.

      • Thanks George
        You’re probably right; the jury’s still out on LDL cholesterol per se.
        Triglycerides were lowered and HDL cholesterol was raised by ezetimibe in IMPROVE-IT. So, we can’t exclude the possibility that this could partly explain the effect of the drug. There’s no proof that it’s only due to lowering of LDL cholesterol.

  7. Yo! Axel
    I am not good at maths….you wrote “. The patients had LDL cholesterol levels between 50-125 mg/dL (1.3-3.2 mmol/L), or between 50-100 mg/dL 1.3-3.2 mmol/L) “…This is ‘black magic’ or incorrect.
    50-100 mg/dL should read 1.3-2.56 mmol/L.

    An RCT separating ‘Primary care’ and ‘Secondary care’ would be useful.
    The anti-inflammatory and Nitric oxide effects of Statins has been known for many years.

    BTW: Why do the Americans keep using the ‘imperial system’ which frustrates the shit out of most people having to convert. Perhaps their general population are not capable.
    There are only 2 other Countries that use this method….Liberia and Myanmar(Burma).

  8. You wrote: “but there were significant reductions in the rate of myocardial infarction (13%, p =0.002), stroke (14%, p=0.052), and ischemic stroke (21%, p=0.008)”

    Are those relative or absolute reductions?

    • There are some errata about the “significant reductions”. From CardioSource:

      Other endpoints including MI (13.1% vs. 14.8%, p = 0.002), stroke (4.2% vs. 4.8%, p = 0.05), ischemic stroke (3.4% vs. 4.1%, p = 0.008), and CV death/MI/stroke (20.4% vs. 22.2%, p = 0.003) were all significantly lower in the ezetimibe/simvastatin arm;

      They also say (just to add the numbers):

      no differences were noted for all-cause mortality (15.4% vs. 15.3%, p = 0.78), CV mortality (6.9% vs. 6.8%, p = 0.99) and need for coronary revascularization (21.8% vs. 23.4%, p = 0.11).

      And as stated by bhrdoc:

      Premature discontinuation was observed in 42% of patients in both arms.

      Color me unimpressed.

      • Thanks Andrés
        Appreciate the details. However I wonder; where’s the errata?

        The relative risk reduction presented was 13% for MI, 14% for stroke and 21% for ischemic stroke. You’ll find it on Medscape/Heartwire and also in Larry Husten’s summary.

        I know you’ll not get exactly the same results (but close) if you calculate from the numbers you provided from CardioSource which is probably because they’re lacking decimals.

      • Oops! The erratum was on my mental syllogisms. I was thinking that to give rise to a meager overall 6.4% reduction in the primary endpoint those relative risk improvements should be lower. My fault. Curious how everyone of them is higher than computed from the absolute risk numbers though. Even more. In the MI case if we take the numbers of absolute risk with one more digit favoring a higher relative risk reduction so 13.05% (instead of 13.1%) versus 14.85% (instead of 14.8%) it gives rise to a relative risk reduction of (14.85-13.05)/14.85=12.12% still lower than 13%.

        It doesn’t put me at ease either that after a median follow up of 6 years they have a non-statistically significant higher mortality on the treatment group both cardiovascular as any-cause. How this trial can be seen as a success is beyond me. I can imagine now how it is going to be used for selling ezetimibe to anyone in secondary prevention irrespective of their LDL number. JUPITER is already used to statinize anyone irrespective of their CRP.

      • Andrés.
        I agree, difficult to get exactly the same numbers for relative risk reduction when you calculate from the numbers provided. But I guess that’s a minor issue and has some hidden explanations, maybe as I suggested partly because of the decimals.
        I tend to agree with you also on the general result, modest at best. And the lack of effect on mortality is quite disappointing in such a large study.

      • Axel,

        “I tend to agree with you also on the general result, modest at best. And the lack of effect on mortality is quite disappointing in such a large study.”

        Well, since the statin-only arm had very low LDL, too, it’s kinda to be expected that the differences wouldn’t be that dramatic.

  9. George,

    that was a bizarre conclusion. The people who have low LDL in Western socities and have not yet gone through cardiadic event are the ones that are terminally ill or have been put on statins due to preventative measures. The latter group have often received statins due to chest pains. Cholesterol levels are quickly reduced, but the existing atheroma plaque is not. Moreover, as these Swedish researchers observered, LDL levels decline temporarily after MI. This trial was about people who were assigned to the trial immeadiately after their first cardiadic event.

    Erik, right on. Whether LDL predicts risk is irrelevant at this point since its the causal factor anyway. We need to be able to differentiate between risk predictors and causal factors when necessary. F.ex. HDL-C predicts risk but it cannot be modulated as such to alter the prognosis. It’s like watching nascar, a risk predictor for premature death due to CHD.

    • HDL predicts risk and can be modulated to improve prognosis by Mediterranean diet, with better results than lowering LDL by drugs. Probably other diets too – LCHF is pretty good for moving this marker too, but hasn’t been trialed for CHD like Med has.
      Drugs that move HDL are hopeless (this includes high intakes of alcohol), but drugs, for obvious reasons, have never proved anything about diet.

      If low LDL is reverse causation by these patients being aware of ill health and trying to improve prognosis before the event, were their other risk factors also better than those of the general healthy population? Was CRP lower, or HbA1c, or TG/HDL etc?

      It would be funny if these supposed health improvements only shifted one marker.
      And funny that this trial only looked at one old-fashioned marker in its crudest measurement – did nothing else shift?

  10. So lest check in with who did this research in the first place becasue one thing there is categorical evidence for is overwhelming disotrtion and bias from drug companies making it fundamentally untrustworthy unelsss we can fine tooth comb the original data – the drop out rate alone makes this suspiscious to say the least.
    But then even if we are being charitable what do we ultimately find – no evidence of reduction in mortality – so what’s the point? Your death certificat date stays the same. So maybe you will have reduced MI’s but you also will have increased diabetes, memory loss probably heart failure and maybe cancer. Oh and the drug comapaniese will have made a few more zillions.

  11. “The lipid hypothesis implies that cholesterol, particularly LDL-cholesterol plays a key role in causing atherosclerosis and coronary heart disease. It means that when it comes to heart disease, measures that elevate blood levels of LDL-cholesterol are usually bad and measures that lower it are good.”

    The use of statins to lower cholesterol may not seem to cause harm beyond the usual side effects. However, the still oft-repeated warning to replace saturated fats with polyunsaturated oils causes considerable harm. Excerpts from page 18 of “Health Preserver : Defining the Versatility of Vitamin E” by Wilfred Shute, MD, 1977:

    “I … believe that the harm done by those who advocate increasing the proportion of polyunsaturated fats and restricting animal fats in the diet is incalculable…The suggestion that these low-cholesterol products contain enough vitamin E to protect the body from the polyunsaturated fat menace is not true…A most unfortunate aspect of this problem is the American Heart Association’s activity in pushing the idea of restricting animal fats, along with the addition of polyunsaturates as a protection against arteriosclerosis and heart disease in children, even in the new-born…More and more pediatricians are placing babies on formulas high in polyunsaturated fats without supplementing these formulas with additional vitamin E, something which must always be done.”

    Now, here’s what the Harvard School of Public Health says about the matter:

    “There has been much confusion and sensational headlines about the role of different types of fat in CHD,” said Frank Hu, senior author and professor of nutrition and epidemiology at Harvard School of Public Health. “Randomized clinical trials have shown that replacing saturated fat with polyunsaturated fat reduces total and LDL cholesterol. And our comprehensive meta-analysis provides clear evidence to support the benefits of consuming polyunsaturated fat as a replacement for saturated fat.”

    Clear evidence? Perhaps so if lowering risk for heart attack is the only consideration. But omega-6 linoleic acid has other effects. Quoting Stephen Gaulin:

    “Back in the 1960s, in the middle of the cardiovascular disease epidemic, people got the idea that saturated fats were bad and polyunsaturated fats were good,”… “That’s one reason margarine became so popular. But the polyunsaturated fats that were increased were the ones with omega-6, not omega-3. So our message is that not only is it advisable to increase omega 3 intake, it’s highly advisable to decrease omega-6 — the very fats that in the 1960s and ’70s we were told we should be eating more of. ”Gaulin added that mayonnaise is, in general, the most omega-6-laden food in the average person’s refrigerator. “If you have too much of one — omega-6 — and too little of the other — omega 3 — you’re going to end up paying a price cognitively,” he said.

    For more on the cognitive disadvantages of replacing saturated fats with polyunsaturated seed oils read this blog post by Australian lawyer David Gillespie entitled “Margarine makes your kids stupid enough to eat margarine.”

    I can’t help but wonder if eating margarine doesn’t make scientists stupid as well.

    • Davebrown9,

      I am an “Ornish-head” and follow the Ornish diet to keep my LDL within physiologic range, <1,8mmol (69,6mg/dl). I prefer tahini on my rye bread. However, Ido consume regularly but sparingly canola-oil based margarine. Given the 100 years worth progressive research into diet-heart, I think that skipping the animal fats is a no brainer.

      Age-adjusted CHD mortality fell around 80% in both Finland and New Zealand starting from the 1970s. This was explained by the change in mean cholesterol levels over few decades (the change occurred prior the wide-scale adoption of statins). The most important dietary change that took place at the time was switch from butter to margarine which resulted in the decline cholesterol. Similar, albeit less radical changes in CHD mortality throughout the developed world. In US mean cholesterol levels fell from the 240s to 200s in few decades followed by dramatic decline in the intake of butter and eggs. Age-adjusted CHD mortality in US was halved.

      Stamler J, Greenland P, Van Horn L, Grundy SM. Dietary cholesterol, serum cholesterol, and risks of cardiovascular and noncardiovascular diseases. Am J Clin Nutr. 1998 Mar;67(3):488-92.

      • Butter was 4.8% of fat calories in the USA in 1954, according to Ancel Keys – unlikely to be a big contributor to CHD there.
        In New Zealand almost all butter (a lot) was eaten on white bread or in biscuits and cakes (4oz butter, 4oz sugar, 4oz flour and an egg).
        This is a bit different from cooking spinach in butter or putting sauce bearnaise on a steak. It’s even different from bulletproof coffee.

      • @Richard

        While CHD mortality has dropped substantially in many developed countries, morbidity remains extremely high compared to the pre-omega-6 industrial seed oil era. I suspect that the Ornish low-fat approach restricts the absolute amount of omega-6 to where it furnishes substantial protection from CHD.

        As for me, I have been consuming between 2 and three pounds of butter a week for several decades; this in addition to full-fat dairy products of all sorts including two gallons of raw, whole milk per week. I’m considerably healthier than I was 7 years ago when I was still consuming peanut butter sandwiches for lunch almost daily during the work week. Replacing peanut butter with thin-sliced turkey or ham was the only dietary change I made.

        Whatever promotes sound health is what a person ought to do so I have no problem with the low-fat approach if it is well tolerated. Some people I know cannot tolerate a high-fat diet.

  12. Thanks for the artcile, and keeping us up to date 🙂

    Did I understood this correctly:

    “this leads to an increased availability of LDL-receptors on the surface of cells leading to increased uptake of cholesterol by liver cells and therefore lowering of blood cholesterol levels”

    So in practise does this then also mean that number of LDL particles in the blood stream is reduced by ezetimibe ?

    All in all, effects are still quite limited as long as there isn’t any reduction death rates.

    Is there any known remarkable side effects known for ezetimibe ?

    • You’re welcome Someone
      Ezetimibe seemed to be well tolerated in this particular study.
      You’re right. In theory ezetimibe should lower LDL particle number as should statins. However, ezitimibe is a very weak LDL-lowering agent compared to statins.

  13. Thanks for all your work Doc Axel.

    Which has the stronger anti-inflammatory effects, statins or ezetimibe?

    Could too much dose intake of ezetimibe, possibly lead to increased internal bleeding?

    • Stephen
      Haven’t heard or read that internal bleedings are a problem with ezetimibe.
      Statins have an anti-inflammatory effect. To my knowledge ezetimibe has no such effect.

  14. Doc,

    will ezetemibe be back in your arsenal of drugs after the positive trial that demonstrated a benefit even when compared to a group with already very low cholesterol levels?


    what do you mean with morbity?

    A significant decline in serum cholesterol, and mortality from cardiovascular disease and all-causes in former communist nations of Eastern Europe took place beginning in the early 1990’s when the communist subsidies on meat and animal fats were abolished after the breakup of the Soviet Union. Likewise, the significant decline in serum cholesterol, and mortality from cardiovascular disease and all-causes in the pre and early statin period of the second half of the century in developed nations throughout Western Europe, North America and Australasia is partly explained as a result of successful government policies that emphasized dietary changes, particularly a decreased intake of saturated animal fat. One of the best examples is Finland which experienced the most rapid decline of coronary mortality in the world, which was predominantly explained by a significant decline in serum cholesterol as the result of a large reduction in saturated animal fat. F.ex Finnish males live 10 years longer today compared to what they did in the 1970s.

    Its not fun to be get a cardiadic event in your 50s and potentially live as a paralyzed impotent for the rest of your life as did hundreds of Doc’s fellow country mates in Iceland few decades ago. Cholesterol lowering works. Veggie fats have saved millions of lives.

    • Richard.
      Ezetimibe has always been in my arsenal of drugs. We use it primarily in patients with severe hypercholesterolemia who don’t reach targets by statins alone.

    • Richard,

      Morbidity refers to the incidence of disease. Mortality refers to the death rate from disease. Mortality in Finland declined dramatically but both CHD morbidity and mortality remain unacceptably high as is true in all developed countries.

      To me it’s clear that the novel dietary factor that contributes most to morbidity and mortality is still being ignored. Interestingly, the edible oils industry has taken steps to reduce the omega-6 linoleic acid content of the food supply so as to make frying oils last longer. Excerpt: “High oleic oil is any oil that is high in monounsaturated fats. Olive and canola oil are naturally high in monounsaturated fat, but they are also high in polyunsaturated fats which mean they are not very shelf-stable. In recent years, scientists have developed sunflower (and other) oils that are bred to be high in monounsaturated fats and low in polyunsaturated fats so they can be used in products that need to be shelf-stable.”

      We should be seeing a dramatic decrease in CHD morbidity and mortality over the next two decades in all parts of the World as the new products replace the high omega-6 versions.

      • The latest Finnish study (KIHD) shows no correlation between saturated fat and CHD mortality or events even in a population where some people are eating a LOT of it (mostly dairy).
        But there is a protective effect of PUFA (canola and fish) whether it replaces SFA or carbohydrate in the diet. Up to about 5%E, not much difference above that. Easy to get 5% PUFA on a LCHF diet.

      • George,

        it’s probably not surprising that SFA isn’t problematic when considered in isolation, that is, without any regard on what it’s replacing? The whole point is – and has been – that PUFA is the better option.

      • True Mie,
        PUFA is the better option for CHD mortality (at least in epidemiology, not so much in RCTs), but makes no difference to overall mortality (meaning that if it prevents CHD deaths, it is also causing non-CHD deaths).
        However, PUFA is also the better option to carbohydrates; in fact, replacing 5%E from carbohydrate with 5%E from linoleic acid is slightly better than replacing SFA.

        Generally anyone switching to a LCHF diet is going to get more PUFA and less carbohydrate. If the goal is to treat CHD, it currently looks as if a Mediterranean type LCHF diet is best.

      • “PUFA is the better option for CHD mortality (at least in epidemiology, not so much in RCTs), but makes no difference to overall mortality (meaning that if it prevents CHD deaths, it is also causing non-CHD deaths).”

        First of all, there’s trial evidence of better CHD health too, not necessarily mortality-specific as e.g. the effects of exhange on e.g. lipid levels aren’t necessarily that high, trial populations heterogenous etc.

        Second of all, if tot. mortality doesn’t change, it doesn’t (necessarily) mean that PUFA causes non-CHD. This needs to be shown specifically, it cannot be deduced from trials powered mainly in detecting CVD endpoints.

      • True, that would need to be tested separately. The current finding is that the whole policy of switching SFA for PUFA has this effect, not the individual components.
        The outlier study, Mann et al. 1997, suggests that it is in fact the restriction of animal fat, cholesterol, and SFA that is strongly (and specifically) associated with increased non-CHD mortality, rather than the PUFA.
        This is about mortality not events, the effect is strong and significant, and death has been found to be diagnosed with more certainty than other endpoints,

  15. George,

    there are several issues why it is hard to show an effect on SFA to CHD within a homogenous population where everyone eats a high fat diet. One of them is inter-individual variation (not to be consumed with intra-individual variation which refers to regression dilution bias).

    Interindividual varition; individual differences in the response to the dose

    The differences in fat intake are usually minor in homogeneous cultures. However, the cholesterol levels in every population follows a normal distribution: a same diet will lead to wide variety of cholesterol levels between individuals, even though everything else is kept constant. If the differences in the fat intake are small in a given population, it cannot explain the large differences in serum cholesterol levels in a model. Jacobs et al (1) showed mathematically that null association is expected between diet and serum cholesterol levels in cross-sectional population studies (even when there exist cause and effect): despite the null associations in such studies, the intake of saturated fats is one of primary determinants of serum cholesterol levels of an individual (Pedersen et al 2011).

    The key to understand the SFA puzzle lies in these 2 articles.

    1) The importance of reducing SFA to limit CHD

    2) Response to Hoenselaar from Pedersen et al.

    The authors make the following point: SFA elevate LDL, and in animal and human models elevated LDL alone promotes chronic disease even in the absence of other risk factors. It does not matter what is the mechanism used to elevate the LDL cholesterol. It should follow that the same is true for humans, any increase in LDL is inherently harmful, and we don’t have evidence on humans to would show otherwise. The second point they make is that population with very low cholesterol levels though out the life do not have CHD; they refer to Campbell’s ecologic evidence from rural China. In addition they refer to the well established IMPACT models. Third, they point to some key pitfalls of our prospective cohort studies:

    “These observations, replicated in many other countries, should not be ignored even if meta-analyses of prospective cohort studies suggest no independent associations of SFA intake with CHD risk(2). The null results of the latter studies(2) probably reflect measurement error, residual confounding, over-adjustment by covariates on the causal pathway and large variations in plasma cholesterol compared to variations in intake of dietary fat(3,12 – 15). The role of SFA risks may also be overlooked, given the strong emphasis on TFA(16) and the incorrect pro- position that the CHD epidemic in affluent societies has been primarily linked to a high consumption of TFA(17)”

    • To believe in an effect of SFA, considered monotonically and not as a component of an unhealthy food (e.g. a donut), even when powerful data sets don’t show it, is the opposite of causality.
      The presumption of causality in epidemiology is based – among other things – on the STRENGTH of the association – in part because error will exist (and not just in the one direction that you’d like it to exist).
      There is also the question of dose-response. If people eating twice as much SFA in a homogenous population (this was the difference between upper and lower quartiles in the KIHD paper) don’t have a higher rate of CHD because the whole population is exposed to SFA, there’s no dose-response. If SFA elevates the LDL fraction that elevates CHD risk, eating twice as much SFA really has to increase risk, so why doesn’t it?
      Otherwise, you’re saying only low-fat vegans are safe (and I’d say yes, the evidence suggests that you can somewhat trade CHD risk for other diseases and causes of death, but not longer life expectancy, if you go to extremes), in which case anyone who opts to eat normally still has no reason to worry about SFA – in for a penny, in for a pound.

      Pedersen et al. cite the New Zealand experience – like Finland, there were huge changes in New Zealand life, not just diet, during those years. The food supply in the 70s was limited and had identified deficiencies such as selenium (corrected in the 1980s in both Finland and New Zealand). Greater choice and more imports and fortification probably did improve health, and increased PUFA – from a very low level to 5%E – probably played a role in reducing CHD. The main benefit was from huge reductions in smoking, as well as similar reductions in air pollution and in problematic industrial and agricultural chemical use. You can see strong associations for all these things (PUFA, smoking, air pollution, persistent industrial and agricultural chemicals), unlike SFA. These associations aren’t swallowed up by error or confounding factors – the benefits of PUFA or nuts are very consistent in the same studies that find no effect of SFA.

      We have to get used to the idea that there are subtypes of LDL and HDL, and that while the amount of LDL and HDL is influenced by fats, the subtypes are overwhelmingly selected by carbohydrate, and the most atherogenic subtypes are selected for by hyperinsulinaemia,
      This is why a lot of butter used in baking or on bread is associated with CHD in ecological historical narratives like the New Zealand one, yet when you isolate SFA and carbohydrate they are independently neutral.
      It’s amazing that Pedersen, writing in 2011, makes no reference to lipoprotein subtypes.

      • George,

        there’s not a single serious lipid researchers who keeps touting about lipoprotein sub-types. The sub-fraction talk comes from the low-carb echo-chamber aided by those wishing to selling expensive lab services to ill-informed.

        Unfortunately, I cannot reply in you properly, have a work to do:

        “All of this was exciting because, as cardiac geneticist Sekar Kathiresan says, “If there’s one thing in medicine that’s certain, it’s that LDL causes heart attack”.

        The 7CS study revealed that if you 100 people ingesting the same diet, the median cholesterol will be higher in the population that consume more SFA (inter-individual variation is eliminated with the simple use medians) and obviously have higher rate of CHD.

        The Japanese baseline stats showed that cigarette exposure was higher than in any other country, yet CHD mortality was 1/10 of the level in US (blood-pressure was as high). The only plausible expiation for this is cumulative exposure to very low cholesterol levels. This notion is supported with genetic data. People with life-time low cholesterol do not get CHD despite being smokers, over-weight, having glucose problems, etc.

  16. So here we have a decision to make.
    We can dismiss research into lipoprotein subclasses as having no bearing on one’s risk of heart disease because they are costly to test (even though LDL/HDL and TG/HDL are fairly good cheap proxies) and because they would imply a benefit for LCHF (which we will still see on a standard obsolescent lipid panel anyway).
    Or, we can throw away everything but the LDL and total cholesterol readings on the vintage lipid panel and while we’re at it ignore the protective associations for high LDL in older people who don’t have CAD or diabetes.
    It seems life expectancy in China wasn’t that great anyway. The long–living countries here are the high fat, high meat ones.

    • Yes,

      the longest-living people seem to all come within cultures with the highest sugar intake. These countries also have the most dense network of McDonalds outlets. Do you think there’s something else going on, such as high infant mortality in less developed nations? Not to speak of state-of-the art medicine. Life saving drugs such as Ezetimibe are widely available in developed nations.

      I’ve never really understood the infatuation with mortality within the low-carb echo-chamber. Doesn’t the quality of life mean anything? Speaking of mortality and diet you should check the blue zones. As a rule people in these cultures eat diametrically the opposite to paleo/lchf. F.ex the people in Ikaria sustain on legumes, potatoes and wheat bread. Very little meat is consumed.

      One cannot interpret the data so that old people with high blood-pressure, LDL cholesterol and body weight are protected. The only reasonable conclusion from this snap-shot data of very old people is that rapid decline of blood-pressure, cholesterol and BMI in an old age is a causal marker of rapidly declining health due to aggressive chronic disease condition.

      • The prospective benefits of high LDL in people without pre-existing disease kick in when they are in their mid-50s in some studies.
        Does fast food prolong life? Well, if it’s low in saturated fat and high in sugar and PUFA, then cholesterol would be lower, wouldn’t it?
        I don’t discount the possibility that lowering LDL by eating a plant-based diet protects against CHD, if your gut will tolerate that sort of food and you can derive nutrition from it. The problem is twofold. One, this is probably not a diet that helps prevent any other disease. Two, most people with CHD don’t have guts in such good shape that they can ferment large amounts of vegetable foods and derive adequate nourishment from them.
        If you look at the traditional Chinese diet, what are they eating? Lots of green vegetables – cooked, not raw. Polished rice – easy to digest. And nose to tail animal food. They may eat less meat, and less protein, than the average westerner, but they definitely eat more liver, tripe, chicken feet, duckbills, fishheads, beef tendons and so on. If you want to eat a low fat, moderate protein diet while avoiding malnutrition and metabolic disease, that looks like the best way to do it.

    • George,

      all LDL subtypes are atherogenic in large quantities. And they are affected by different fats, too, BTW.

      And I don’t think anyone is suggesting throwing away everything in risk prediction but LDL and/or tot. cholesterol, especially as there are quite a few people (met. syndrome/DM2 patients) who often have discordance in the sense that despite lower LDL count – compared to people without the aforementioned conditions – they still have higher risk of CVD.

      • Right,
        I am of course responding to RichardOFL’s reductio ad absurdum rather than making a position statement.
        It is rather well known that the Ornish diet lowers HDL.
        If RichardORL thinks that LDL is the only marker we should bother about – well, he would wouldn’t he? No doubt there is an echo chamber somewhere for that.
        The HDL effect seems to be depended on the functionality of HDL in reverse transport
        “HDL has numerous antiatherosclerotic actions that are not readily reflected by HDL cholesterol levels. A key function of HDL is to promote reverse cholesterol transport from the periphery to the liver, and the critical initial step in reverse cholesterol transport is cholesterol efflux from macrophages to HDL. Macrophage-specific cholesterol efflux capacity has been directly and causally linked to the prevention of atherosclerosis in animal models.”

        What sort of factors improve this HDL functionality? One is the rumenic acid form of CLA, the cis, trans fatty acid found in dairy and ruminant fat, helping to explain the reverse association between full-fat dairy and CHD. To the best of my knowledge there has been no epidemiology that isolates the associations with other sources of ruminant fat.

      • George,

        from what I’ve gathered, HDL functionality isn’t necessarily independent of HDL-C in all cases. That is, the lower the HDL levels, the less functional HDL seems to be. In some cases, at least.

        In addition, certain conditions, e.g. DM2, impair HDL functionality.

        And certain dietary choices,

Leave a Comment

This site uses Akismet to reduce spam. Learn how your comment data is processed.