For 70 years we have been fighting the firefighter. Not the fire.
Cholesterol could be less the perpetrator than the repair crew, trying to make good on what sugar, stress, seed oils, and toxins do to your endothelium. A deep dive through 100 years of science, with verifiable sources, studies, reviews, and guidelines, and my personal story. With the kind of look that ten minutes of consultation rarely leaves room for.
My paternal grandfather died before I was born. Of a heart attack. My maternal grandfather as well. Both. Both supposedly because they had eaten too much meat and too much fat. That was the family truth. And that is exactly why, at 19, in medical school, I started believing what was being told to us up at the front of the lecture hall: high cholesterol? Less butter. Fewer eggs. Less meat. So that you live long enough for your grandchildren to know you. That is how I protected myself. I thought.
In the lecture hall, without a single clean piece of evidence, we were told to recommend that patients eat fewer eggs, less butter, and less meat to protect them from heart attacks.
That is not differentiated medicine. That is the repetition of a mistake we have been making since the 1950s.
I am not fighting the cholesterol value. I am fighting the blanket generalization. A conversation about statins gets better when more than one value is on the table. Apo-B, Lp(a), triglycerides, HbA1c, and fasting insulin give a fuller picture. Feel free to ask your physician about them, many are happy to when you bring it up.
What you are reading now is the answer to a question I could not get out of my head after three years on Paleo: How can it be that what did my whole body so much good is supposed to destroy my heart?
Never stop or reduce a prescribed medication on your own. This applies especially to statins and other cardiovascular drugs. If you want to change something, talk first to the physician who prescribed it.
This text is addressed to healthy adults who want to get their bearings. It is not individual advice and does not replace a medical examination. If any of the following applies to you, every change to diet, supplements, or medication belongs in a conversation with your physician first: familial hypercholesterolemia or another known lipid disorder, a past heart attack or stroke, a stent, a bypass, or known coronary heart disease, diabetes treated with insulin, sulfonylureas, or SGLT2 inhibitors, kidney or liver disease, pregnancy and breastfeeding. Nothing in this text applies to children and adolescents.
One word on a distinction that keeps coming up in this text: primary prevention means people who have never had a cardiovascular event. Secondary prevention means people after a heart attack, stroke, stent, or bypass. For that second group the evidence on cholesterol lowering is considerably clearer, and much in this article expressly does not apply to them.
In an emergency: with chest pain, pressure on the chest, shortness of breath, cold sweat, or pain radiating into arm, jaw, or back, call the emergency number immediately (112 in Europe, 911 in the US). Do not wait and do not book a practice appointment.
Both grandfathers. Both heart attack. Both before I was born.
My paternal grandfather died before my father became a father. My maternal grandfather likewise. Both died before I was born. Both of a heart attack. In my family this counted as proof: we eat too much meat, we eat too much fat, we die young.
I did not want to repeat that. I wanted to grow old. I wanted my grandchildren to know me. So I held to the recommendation that medical school and family history together pressed upon me: less meat, fewer eggs, less butter.
Three years vegan. I was okay. But not good.
I was vegan for three years. I lived it with full conviction. After two years I developed sleep problems. Irritability. A sudden intolerance of legumes, then of gluten. I compensated for years: 30 minutes of breathwork in the morning, 30 minutes of HIIT, then ice baths. With that much discipline, anything works, for a while.
Then Paleo. No grains, no legumes, no dairy, no nightshades.
What was left: meat, eggs, vegetables, fruit. I deliberately left out fish because of the heavy-metal burden, especially methylmercury. About 250 grams of meat and at least five eggs a day. At first I thought it was crazy.
After three months I felt better than ever before. It kept going. Better and better, even though I had not felt bad before. With Paleo I did not have to compensate anymore. Twice the energy, and balanced.
After nine months the white spots on my fingernails were gone, and my beard grew in properly for the first time. Whether that was down to the diet I cannot prove, I was also simply a few years older by then. What I am describing here is my own observation, not proof of an effect, and certainly not an expectation for you. My relationship to stress also changed in that period.
After a year I could tolerate legumes and gluten again.
At 28 I felt fitter than at 18. And at 18 I had thought I was indestructible. Right at that moment came the question that would not let me go: can something that does my whole body so much good destroy my heart? It makes no sense that my whole body feels noticeably better and then I die earlier of a heart attack. Something is missing in the story the university told me.
That question pushed me to systematically work through every relevant cholesterol study between 1913 and 2020. What I found is this article. I felt deceived. I was angry. We keep making the same mistakes in history.
Without cholesterol almost nothing works in your body. No cell membrane, no steroid hormone, no vitamin D, no bile.
Before we talk about risk, we have to talk about function. Because function has almost dropped out of every consultation.
Cholesterol is the building block of your cell membrane. Every single cell in you, from a skin cell to a neuron in the frontal cortex, has a membrane that without cholesterol would collapse like a wall without mortar. Cholesterol regulates how fluid or firm that membrane is. Too little of it and signals are no longer cleanly passed on.
Cholesterol is the precursor of every steroid hormone. Testosterone, estrogen, progesterone, cortisol, aldosterone, DHEA. Everything you need to get out of bed in the morning, fall in love, become pregnant, or survive a conflict, begins as a cholesterol molecule in a gland.
Cholesterol is the precursor of vitamin D. The 7-dehydrocholesterol in your skin is converted by sunlight to vitamin D3, which is then activated in liver and kidney. No cholesterol, no vitamin D.
Cholesterol is the basis of your bile acids. Without bile, no fat digestion, no fat-soluble vitamins A, D, E, K.
Cholesterol could be involved in the repair processes of your endothelium. Where your vessel wall develops microscopic tears, cholesterol from LDL particles is deposited into the wall. Whether that is a repair or the beginning of the damage is exactly the disputed question. We will return to that in detail later.
The metabolism in numbers
balances it
Let that sink in for a moment. The larger part of the cholesterol moving through your body every day is built by the liver itself, mainly via the enzyme HMG-CoA reductase. That is exactly the enzyme statins inhibit.
The consequence: if you eat three eggs today, the liver can counter-regulate and build less. If you skip the eggs, it can build more. The arithmetic of “fewer eggs, therefore less cholesterol” does not work out for many people, because the liver adjusts. For some people it does work out, though. There are people whose LDL responds markedly to dietary cholesterol. Which type you are, you only see if you measure.
Cholesterol is not something we eat that harms us. Cholesterol is something we build ourselves, because we need it to survive. When your value is high, the good physician asks: why is your body building so much of it right now? Not: how do we lower it?
And now you know why the old recommendation “eat less butter” delivers less for many people than it promises.
1955, Eisenhower collapses. Ancel Keys delivers the explanation. A whole world believes it.
For you to understand where the myth comes from, we have to go back 70 years. The thesis “saturated fats cause heart attacks” did not emerge from a randomized trial. It grew big in a political moment after a prominent heart attack.
The timeline of cholesterol doctrine
Anitschkow feeds rabbits cholesterol and observes plaques. Rabbits are obligate herbivores. The result is not transferable to humans. The study still becomes the birth certificate of the cholesterol theory.
John Gofman separates lipoproteins by ultracentrifugation and discovers LDL and HDL. Differentiation begins, but it takes decades to reach routine diagnostics.
Eisenhower suffers a severe heart attack. His personal physician Paul Dudley White goes public, issues dietary recommendations. Cholesterol becomes politics.
Ancel Keys starts the Seven Countries Study. How the countries were selected is still discussed controversially today, as is how robust the international comparison was at the time. What is clear: no cause can be derived from a comparison between countries.
The American Heart Association officially recommends, for the first time, replacing saturated fats with polyunsaturated vegetable oils. Worldwide seed oil consumption explodes.
Start of the Minnesota Coronary Experiment. The largest randomized diet experiment in history on the cholesterol hypothesis. The data disappear into a basement for 40 years.
Krauss and Austin describe the LDL subclasses. From now on it is clear: LDL is not equal to LDL. This insight needs another 20 years to reach guidelines, and to this day is not standard in the covered lab panel.
Christopher Ramsden digs out the original Sydney Diet Heart data. In the linoleic-acid group there were more deaths. Published in the BMJ.
Ramsden digs out the Minnesota data. Per 30 mg/dL of cholesterol lowering, 22 percent more deaths. Published in the BMJ.
What follows from this story: the recommendation “less saturated fat” was not born from more knowledge, but from a political moment. And in parallel with the recommendation came a second phenomenon, almost never discussed in practice.
The great fat switch and what came after
After Eisenhower, the United States, then Europe, in the broad population switched from animal fat to industrial seed oils. Corn, soy, safflower, sunflower, canola. Before 1900 the ratio of omega-6 to omega-3 in the Western diet was about 1 to 1. Today it averages around 15 to 1, in the US sometimes 25 to 1.
What has happened since? Cardiovascular mortality in Western countries has fallen substantially since the 1970s. That can have many reasons, from emergency medicine to the decline in smoking to blood pressure treatment. What has risen sharply in the same period is obesity and type 2 diabetes. What is due to what cannot be derived from such time series. Two variables were newly added to the menu in this period: industrial seed oils and refined carbohydrates. Whether they play a role is an open question, not a proven answer.
Is LDL in plaque because plaque is a repair? Or because LDL attacks the wall? That is exactly what is disputed.
Picture the endothelium of your vessels as a smooth, paper-thin foil. A single cell layer thick. It produces the nitric oxide that relaxes your vessels. It prevents blood from clotting where it should not. It regulates what comes in and what goes out.
When this foil tears, microscopically, several times every day, the body needs material to attend to the site. A view that convinces me, but that is not the only one: LDL could be involved in this repair. It transports cholesterol to where the vessel wall is irritated. Over time, plaque forms.
The majority view in lipidology sees this differently. There, the retention of LDL particles counts as the trigger and not as the answer, the so-called response-to-retention model. Both models explain parts of the data. This is not settled. What you are reading here is my reasoned position, not an established fact.
If you lower cholesterol without calming the endothelium, your body could have less material to attend to the daily tears. You may then be fighting the painter, not the hammer that is smashing the wall. If we only lower cholesterol without asking why the damage to the wall is happening, the question of cause remains open.
What are the hammers? The hammers are what can stress your endothelium every day. This is exactly where it is worth going further in the conversation.
What can stress your endothelium day to day
- Hyperglycemia and insulin resistance. High blood sugar spikes generate advanced glycation end products (AGEs) and oxidative stress. This is the most common driver in modern everyday life.
- Linoleic acid from refined seed oils. Sunflower, safflower, soy, corn. It oxidizes easily. Oxidized metabolites (OXLAMs) are found enriched in plaques. They could turn a less reactive LDL into a more oxidation-prone one. Canola does not belong in this group, it is low in linoleic acid and rich in alpha-linolenic acid.
- Chronic stress and constant sympathetic activation. Elevated heart rate, elevated blood pressure, chronically elevated cortisol. This stresses the endothelium mechanically and biochemically.
- Alcohol, regularly. A direct endothelial toxin that promotes lipid peroxidation and insulin resistance.
- Smoking. Direct oxidative stress with every inhalation.
- Heavy metals and mycotoxins. Mercury, cadmium, mold toxins. All are documented endothelial toxins.
- Sleep deprivation. Worsens insulin sensitivity, raises cortisol, reduces HRV.
- Lack of movement. Reduces NO production in the endothelium, promotes insulin resistance.
Mechanism reviewDiNicolantonio & O’Keefe, 2018, Open Heart. The oxidized linoleic acid hypothesis
Published in the peer-reviewed BMJ journal Open Heart. The authors lay out the mechanisms showing that oxidized linoleic acid metabolites are markedly enriched in atherosclerotic plaques. They argue that the driver of atherosclerosis is not cholesterol itself but the oxidative modification linoleic acid from seed oils causes on the LDL particle.
Consequence: linoleic acid reduction is a plausible point of leverage that has so far played hardly any role in German guidelines.
DiNicolantonio JJ, O’Keefe JH. Omega-6 vegetable oils as a driver of coronary heart disease: the oxidized linoleic acid hypothesis. Open Heart. 2018;5(2):e000898. doi.org/10.1136/openhrt-2018-000898
Human study, 8 weeksReaven et al., 1993, J Clin Invest. Olive oil vs. linoleic acid
Subjects with mildly elevated cholesterol received eight weeks of diets rich in oleic acid (olive oil) or linoleic acid (seed oil). The LDL of the olive oil group was significantly more resistant to oxidation. The 18:2 linoleic acid content in LDL correlated directly with the tendency to oxidize. The authors recommended in 1993 to use olive oil instead of PUFAs. It did not become German guidance.
Reaven P, Parthasarathy S, Grasse BJ, et al. Effects of Oleate-Rich and Linoleate-Rich Diets on the Susceptibility of Low Density Lipoprotein to Oxidative Modification. J Clin Invest. 1993;91(2):668–676. doi.org/10.1172/JCI116247
This is the point that matters to me: if we lower cholesterol without addressing these drivers, we are working mainly on the reaction. The cause then stays untouched. Roughly as if one were to send the fire brigade away instead of putting out the fire. That does not mean lowering is pointless. It means it is rarely enough on its own.
First total cholesterol. Then HDL and LDL. Then subclasses. Today Apo-B and Lp(a). In routine insurance-covered care, usually only the first stage arrives.
The history of cholesterol science is a history of progressive differentiation. Each step has weakened or even reversed the previous statement. You have to understand this, otherwise you cannot understand the dispute.
Stage one: total cholesterol
At first only one value could be measured. There was a rough correlation between high total cholesterol and heart attack rate. That was the state in 1950. From this came the recommendation “lower your cholesterol.” Plausible, but too sweeping.
Stage two: HDL and LDL
With John Gofman and ultracentrifugation it became clear: there are at least two main classes. HDL has been considered “good” ever since, LDL “bad.” This is already a simplification. But at least it was clear: high total cholesterol with high HDL means something different from high total cholesterol with low HDL.
Castelli, 1996, Atherosclerosis. What the Framingham data show about the interplay
William P. Castelli was one of the central figures of the Framingham Heart Study. In this overview he makes very clear: there are over 200 known cardiovascular risk factors. The most important are lipid subclasses, triglycerides, blood pressure, and smoking, not total cholesterol alone. Castelli explicitly writes that more than 15 different cholesterol-containing lipoproteins exist and four different triglyceride-rich particles, some of them strongly atherogenic.
He recommends two simple tests: the ratio of total cholesterol to HDL and triglycerides. Only then can it be better judged whether LDL, HDL, or triglyceride-rich particles carry the risk. In the standard covered lab panel this is still not routine.
Castelli WP. Lipids, risk factors and ischaemic heart disease. Atherosclerosis. 1996;124 Suppl:S1–9. doi.org/10.1016/0021-9150(96)05851-0
The most important takeaway from Framingham: in these data the risk did not sit with LDL alone, but with the interplay of LDL, HDL, and triglycerides. Someone with an HDL of 30 and an LDL of 130 is in a different cardiological situation from someone with high HDL and low triglycerides. What that means for you cannot be derived from two numbers. A high HDL does not automatically make a high LDL harmless, especially if there is a family history. And still, in routine care, often only the total cholesterol is reported.
Stage three: LDL has subclasses
Here comes the insight that turns everything upside down. LDL is not a homogeneous substance class. There are at least three measurable subclasses, which differ massively in size, density, and risk.
Phenotype A
Large, light, “fluffy” LDL particles. They release cholesterol cleanly to cells. Low oxidation susceptibility, low entry rate into the vessel wall. Little atherogenic.
Transition
Medium-sized. Atherogenic risk partly raised, especially in combination with insulin resistance and inflammation. On the spectrum between truly harmless and truly dangerous.
Phenotype B
Small, dense, oxidation-prone. Easily penetrates the vessel wall. Stays there long. Strongly atherogenic. Rises under refined carbohydrates, insulin resistance, stress, and especially industrial seed oils.
Vekic et al., 2022, Medicina. Small, dense LDL as an independent risk factor
This systematic overview summarizes: small dense LDL has a higher entry rate into the vessel wall, a longer dwell time there, a markedly higher tendency to oxidize, and a reduced affinity for the LDL receptor. The result is enrichment in the subendothelium and accelerated plaque formation.
The consequence for practice: a high LDL does not automatically equal a high risk. On this evidence, high LDL with a lot of phenotype A means something different from high LDL with a lot of phenotype B. Whoever separates the two sees more.
Vekic J, Zeljkovic A, Cicero AFG, et al. Atherosclerosis Development and Progression: The Role of Atherogenic Small, Dense LDL. Medicina (Kaunas). 2022;58(2):299. doi.org/10.3390/medicina58020299
Mechanism reviewGriffin, 1999, Proceedings of the Nutrition Society
Griffin writes in 1999 that elevated serum cholesterol does not adequately explain coronary risk within a population. In the same breath he holds that LDL nevertheless remains the most atherogenic lipoprotein class, just not because of its cholesterol content but because of the number of small, dense particles.
That is exactly the differentiation I mean. It is from 1999 and is still rarely reflected in the standard covered lab panel.
Griffin BA. Lipoprotein Atherogenicity: An Overview of Current Mechanisms. Proc Nutr Soc. 1999;58(1):163–169. doi.org/10.1079/pns19990022
What drives SD-LDL up?
The most important answer: not the steak. Fast carbohydrates. Bread, juice, sugar, refined starches. They raise triglycerides, and triglycerides raise the share of SD-LDL. Even a low-fat diet that lowers total LDL can raise the SD-LDL share, because it pushes triglycerides up.
Hirano, 2018. Insulin resistance generates the atherogenic lipid profile
This review article shows very clearly: insulin resistance, the precursor of type 2 diabetes, is the driver behind the classic “diabetic fat pattern.” High triglycerides, low HDL, lots of SD-LDL. Exactly the profile found in most Western heart attacks.
Consequence: a cholesterol value without insulin and triglycerides beside it shows only a section of the picture.
Hirano T. Pathophysiology of Diabetic Dyslipidemia. J Atheroscler Thromb. 2018;25(9):771–782. doi.org/10.5551/jat.RV17023
Stage four: Apo-B and Lp(a)
Apo-B stands for apolipoprotein B. Simplified: every atherogenic particle, that is LDL, SD-LDL, VLDL, IDL, Lp(a), carries exactly one Apo-B protein on its surface. When you measure Apo-B, you are counting the number of particles that can really do damage. Not the volume of cholesterol they transport.
Two people can have the same LDL value. One has 800 large, fluffy particles. The other has 1,500 small, dense ones. Same cholesterol value. Different risk, often by a factor of two or three. Apo-B distinguishes this. Total cholesterol does not.
Sniderman, 2013. Apo-B is superior to cholesterol markers
Allan Sniderman, one of the patriarchs of lipoprotein research, shows in this meta-analysis of prospective studies that Apo-B outperforms both non-HDL cholesterol and LDL cholesterol as a risk predictor. Three separate discordance analyses confirm: when Apo-B and LDL-C do not agree, Apo-B is the better predictor. His conclusion: epidemiology without physiology is like firing without aiming.
Sniderman A, Kwiterovich PO. Update on the detection and treatment of atherogenic low-density lipoproteins. Curr Opin Endocrinol Diabetes Obes. 2013;20(2):140–147. doi.org/10.1097/MED.0b013e32835ed9cb
Lp(a), pronounced “lipoprotein little a,” is a largely genetically determined, very atherogenic particle. A high Lp(a) value can go along with a markedly increased risk for early heart attacks, independent of much else. Because the value is mostly genetically set, a single measurement is usually enough. With kidney disease, thyroid problems, marked inflammation, in pregnancy, or on hormone therapy it can change, and then a follow-up measurement is worthwhile. In routine care it is still rarely run.
Kamstrup, 2021, Clinical Chemistry. Lp(a) is its own causal risk factor
Pia Kamstrup of the Copenhagen General Population Study summarizes: Mendelian randomization and large cohorts show a causal contribution of Lp(a) to coronary heart disease, peripheral artery disease, and aortic valve stenosis. High Lp(a) values predict a two- to threefold increased risk, independent of everything else. 10 to 20 percent of the population have high values. Most do not know it.
Kamstrup PR. Lipoprotein(a) and Cardiovascular Disease. Clin Chem. 2021;67(1):154–166. doi.org/10.1093/clinchem/hvaa247
GuidelineESC/EAS Guidelines, 2019. Apo-B and Lp(a) belong in the risk profile
The European cardiologists in the 2019 guidelines recognize Apo-B as an equivalent or better marker than LDL, especially in diabetes, obesity, insulin resistance, or high triglycerides. Lp(a) is included as an independent risk factor. That is the official European recommendation.
In routine covered care it usually stops at total and LDL cholesterol. That is not wrong. It is simply less information than would be possible today.
Important here: extended diagnostics do not replace a therapy that is indicated by guideline. They complement it. If a treatment is indicated, additional lab values do not make it superfluous.
Mach F, Baigent C, Catapano AL, et al. 2019 ESC/EAS Guidelines for the management of dyslipidaemias. Eur Heart J. 2020;41(1):111–188. doi.org/10.1093/eurheartj/ehz455
If your cholesterol value is called “too high,” the next useful question is: “what are my HDL, my Apo-B, my Lp(a), my triglycerides, my HbA1c, and my fasting insulin?” You are allowed to ask that question yourself. It makes the conversation better.
Sydney Diet Heart and Minnesota Coronary Experiment. Two reanalyses that did not show what was expected.
The most important randomized controlled trials ever conducted on the cholesterol hypothesis lay in drawers for decades. Only Christopher Ramsden of the US National Institutes of Health dug them out again.
RCT, n=458Sydney Diet Heart Study (Ramsden 2013)
458 men, all after a heart attack. One group replaces saturated animal fats with safflower oil and margarine. The other group stays with normal diet. Follow-up over several years.
Result on reanalysis: in the linoleic acid group more people died. 17.6 percent versus 11.8 percent all-cause mortality. Hazard ratio 1.62, confidence interval 1.00 to 2.64, p equals 0.05. The lower bound sits exactly on 1.00, so the result stands right on the significance threshold. From cardiovascular disease 17.2 to 11.0 percent.
Important framing: this was secondary prevention in men aged 30 to 59 after a heart attack, not the readership of this article. The meta-analysis updated by the same authors showed only non-significant trends, for coronary death 1.33 with an interval of 0.99 to 1.79.
Ramsden CE, Zamora D, Leelarthaepin B, et al. Use of dietary linoleic acid for secondary prevention of coronary heart disease and death. BMJ. 2013;346:e8707. doi.org/10.1136/bmj.e8707
RCT, n=9,423Minnesota Coronary Experiment (Ramsden 2016)
A double-blind randomized study in 9,423 people. Test: if saturated fat is replaced with corn oil, does cholesterol drop? Does it protect from cardiac death?
Result: cholesterol fell by 13.8 percent. No protection from cardiac death appeared. In an adjusted regression analysis within the study, every 30 mg/dL of lowering went along with a 22 percent higher mortality risk.
Important framing: the direct comparison between groups, that is the actual randomized part, showed no difference in mortality between the groups. The 22 percent come from an observational analysis inside the study, not from the randomization result. Only 2,355 of the 9,423 participants received the diet for a year or longer. The accompanying meta-analysis found no significant difference for all-cause mortality (1.07; 0.90 to 1.27). The original data sat unused for 40 years.
Ramsden CE, Zamora D, Majchrzak-Hong S, et al. Re-evaluation of the traditional diet-heart hypothesis: Minnesota Coronary Experiment. BMJ. 2016;353:i1246. doi.org/10.1136/bmj.i1246
RCT, n=26 (crossover)Berry et al., 1991. Jerusalem Nutrition Study
26 young men received in crossover design twelve weeks each of a diet rich in monounsaturated fats (olive oil) or polyunsaturated fats (typical seed oils). Both lowered total cholesterol. PUFA more strongly (16 percent). MUFA less (10 percent).
The interesting observation: the LDL of the PUFA group was more oxidation-prone, measured by TBARS values. More cholesterol lowering, but a more oxidation-prone LDL. That is a conceivable mechanism, no more than that. 26 healthy young men over twelve weeks with a surrogate marker do not explain mortality data.
Berry EM, Eisenberg S, Haratz D, et al. Effects of Diets Rich in Monounsaturated Fatty Acids: the Jerusalem Nutrition Study. Am J Clin Nutr. 1991;53(4):899–907. doi.org/10.1093/ajcn/53.4.899
“Vegan lowers LDL, therefore vegan protects from heart attack.” This bridge must not be built.
If I had to take one single thing from 100 years of cholesterol science, it would be this: there is a logical leap that has stood in German textbooks for decades and is not tenable as such.
Step one: LDL correlates with cardiovascular disease. True for the LDL species overall. True especially for SD-LDL.
Step two: vegan diet lowers LDL. True on average.
Step three: therefore vegan diet protects from heart attack. Scientifically, you may not take this step. It is not proven. It is a bridge no one should have built.
Dinu et al., 2017. What looks better in the vegetarian data
This large overview evaluated 86 cross-sectional studies and 10 prospective cohorts. Vegetarian diet lowers BMI, total cholesterol, LDL, and glucose. It is associated with about 25 percent less ischemic heart disease in observational data.
It also found a lower cancer incidence, about 8 percent, and about 15 percent in vegans. For all-cause mortality, cancer mortality, and stroke, no significant difference was shown. The picture is therefore mixed, and it comes from observational data.
Dinu M, Abbate R, Gensini GF, Casini A, Sofi F. Vegetarian, vegan diets and multiple health outcomes. Crit Rev Food Sci Nutr. 2017;57(17):3640–3649. doi.org/10.1080/10408398.2016.1138447
Beyond that: plant-based diets can raise the share of SD-LDL, even when total LDL falls. They can deliver lectins, phytates, and oxalates that for some people can act inflammatorily. They can lead to deficiencies in B12, iron, zinc, and DHA, which are themselves cardiovascularly relevant. My own body felt this very clearly in three years vegan.
Lowering a laboratory value is not proof of a longer life. The goal is not low cholesterol. The goal is intact endothelium. These two sentences sound similar. But they are completely different.
There is a great deal of observational research on meat. And it is contradictory.
Meat has been part of human nutrition for thousands of years. A large review by the NutriRECS consortium from 2019 concluded that the certainty of evidence for the recommendation to eat less red meat is low. At the same time, the WHO cancer research agency classifies processed meat as carcinogenic and red meat as probably carcinogenic. Both stand side by side, and both belong in an honest answer.
What does allow a statement is differentiation. Meat is not equal to meat. The following factors decide whether meat is associated with disease in a dataset:
What is often really behind “meat is unhealthy”
- Husbandry and feeding. Animals from mass production fed soy and corn have an omega-6 to omega-3 ratio of about 15:1 or more. Pasture meat from a cow that ate grass sits at about 2:1.
- Processing. Sausage, salami, hot dogs are not the same as a steak. Curing salts, nitrates, smoking processes generate other compounds that have their own effects.
- Combination. Whoever eats meat with fries, beer, and cigarettes lives differently from someone eating pasture meat with vegetables and olive oil. Observational studies can rarely separate this cleanly.
- Concomitant factors. People who eat lots of processed meat also tend, in many studies, to eat fewer vegetables, more sugar, more alcohol, are more often overweight, and move less. “Meat” is then a marker for an entire lifestyle.
- Total calories and insulin resistance. Large caloric intake with insulinogenic effect creates the metabolic state in which the endothelium is chronically irritated.
My personal position: quality and context make the difference. The question is always: how much, combined with what, in what context, with what insulin resistance. When these factors are taken into account, the picture becomes considerably less uniform than the headlines suggest. With processed meat, though, I am cautious myself, since that is where the evidence is clearest.
A large meta-analysis from 2022 (Zhao et al., Circulation) found that every additional daily egg portion of 50 g was associated with an increased cardiovascular risk. The same paper also found that higher serum total cholesterol went along with higher cardiovascular mortality. That speaks against my reading, and I am therefore leaving it in. What remains open: these are observational data, the paper’s own cohort was a cohort of smokers, and the effect sizes are small. The dispute is not decided.
We are not all alike. And that is exactly what makes every blanket dietary recommendation questionable.
The mitochondrial researcher Douglas Wallace has shown something exciting: our mitochondrial DNA has, depending on where our ancestors lived for thousands of years, adapted to the food available there.
Anyone with a family lineage from the far north tends to carry mitochondrial haplotypes that metabolize fat very efficiently. Inuit, Sami, and northern European lines show clear adaptations to fat-rich, low-carbohydrate nutrition over many generations.
Anyone from the equatorial belt tends to be better adapted to plant carbohydrates, because photosynthesis runs all year there and plants are available year-round.
Enzymatically too there is variance. How well someone digests meat differs a great deal between individuals, something the biochemist Roger Williams described back in the 1950s in his classic Biochemical Individuality. Robust modern figures on this are lacking. Even in populations with a very fat-rich diet, such as the Maasai, the picture is mixed: they are often metabolically healthy, and autopsy studies nevertheless found atherosclerosis. The Hadza in turn eat lots of honey and plants and are also metabolically healthy. Not one diet fits all.
So you do not misunderstand me: LDL can be a causal factor. But not every LDL. And not at every age.
An honest cholesterol discussion has to show both sides. Here are the most important studies that challenge or differentiate my position.
Ference et al., 2017, European Heart Journal. LDL is causal in ASCVD
This consensus statement of the European Atherosclerosis Society summarizes more than 200 cohort studies, Mendelian randomization studies, and randomized trials with over 2 million participants and over 150,000 events. The result: LDL has a causal effect on atherosclerotic cardiovascular disease, log-linearly dependent on level and duration.
That is the solid counter-position. It is real. What differentiation does allow: this consensus statement concerns the LDL species, not the individual person with phenotype A, good HDL, low triglycerides, and clean Apo-B. For that person, risk cannot be derived from the LDL number alone.
Ference BA, Ginsberg HN, Graham I, et al. Low-density lipoproteins cause atherosclerotic cardiovascular disease. Eur Heart J. 2017;38(32):2459–2472. doi.org/10.1093/eurheartj/ehx144
Ravnskov et al., 2016, BMJ Open. In the elderly, LDL-C is often inversely associated with mortality
This systematic overview evaluated 19 cohort studies in people over 60. In the 16 cohorts in which all-cause mortality was recorded, people with higher LDL cholesterol lived as long or longer, which arithmetically covered 92 percent of the subjects. The association was statistically significant in 14 cohorts.
The authors’ conclusion: the cholesterol hypothesis is not tenable in this form for the elderly, and routine LDL lowering in this age group should be reassessed.
What you should know while reading: this paper and the statin article by Okuyama cited further below come from the same group of authors, who have positioned themselves critically towards the cholesterol hypothesis for years. That does not invalidate their arguments. But they are not two independent pieces of evidence.
Ravnskov U, Diamond DM, Hama R, et al. Lack of an association or an inverse association between low-density-lipoprotein cholesterol and mortality in the elderly. BMJ Open. 2016;6(6):e010401. doi.org/10.1136/bmjopen-2015-010401
This is where the KETO-CTA paper by Soto-Mota and colleagues from JACC Advances 2025 originally stood. That study was retracted after publication at the request of the authors and the editors, because concerns about the methodology affect the reliability of the data (retraction notice, JACC Adv. 2026;5(5):102824). I have therefore removed it here.
The question of whether there are subgroups in which the LDL model does not apply remains open. It cannot be answered with this paper. I am leaving the note in rather than deleting it quietly, because you should know what I used to lean on.
There is very strong evidence that LDL as a species contributes causally to atherosclerosis. There is also strong evidence that this does not apply equally to every person, in every age, in every metabolic context. Both must be allowed to hold at the same time. It is exactly this side-by-side that tends to get lost in a short consultation.
Statins work. And statins do not work. Both are true, depending on whom you give them to.
I am not against statins. In people with familial hypercholesterolemia, after a heart attack, with documented plaque and several risk factors, the evidence clearly favours them. Where I look more closely is primary prevention in young, otherwise healthy people with isolated high LDL and an otherwise unremarkable profile. There the benefit depends strongly on the baseline risk.
And once more, because it is the most important sentence in this chapter: if you take a statin, please never stop it on your own and do not reduce it on your own initiative. Talk first to the physician who prescribed it. Especially after a heart attack, a stent, or a bypass, stopping on your own can be dangerous.
Cai et al., 2021, BMJ. Statins in primary prevention
A systematic review from Oxford with network and dose-response meta-analysis. It found somewhat more self-reported muscle complaints, liver and kidney abnormalities, and eye conditions under statins. For clinically confirmed muscle disorders and for diabetes it found explicitly no association.
The authors emphasize at the same time that these risks are small in absolute terms and do not outweigh the benefit. Their conclusion is that the benefit-to-harm balance of statins is generally favourable. What remains important is that the absolute benefit depends on how high your baseline risk is.
Cai T, Abel L, Langford O, et al. Associations between statins and adverse events in primary prevention. BMJ. 2021;374:n1537. doi.org/10.1136/bmj.n1537
Sattar et al., 2010, Lancet. Statins raise diabetes risk
9 percent increased risk of new-onset diabetes under statins, in absolute terms one additional diabetes case per 255 people over four years. The authors state explicitly that clinical practice at moderate or high risk should not change, because the protection from heart attack outweighs this. Notably, the diabetes risk was highest in trials with older participants. In young people with low baseline risk the benefit-risk calculation is a different one overall, simply because there is less to gain there.
Sattar N, Preiss D, Murray HM, et al. Statins and risk of incident diabetes. Lancet. 2010;375(9716):735–742. doi.org/10.1016/S0140-6736(09)61965-6
Hypothesis, no original dataOkuyama et al., 2015. A hypothesis about statins and mitochondria
A group of authors around Okuyama published in 2015 the consideration that statins could also have unwanted effects via coenzyme Q10, heme A, vitamin K2, and selenoproteins. This is a perspective article without original data and strongly contested in the field. It comes from the same group of authors as the Ravnskov review cited further above.
I mention the consideration because it is published and pharmacologically debatable. It is not proof of an effect. Important: if you take a statin, please never stop it on your own. Talk first to the physician who prescribed it.
Okuyama H, Langsjoen PH, Hamazaki T, et al. Statins Stimulate Atherosclerosis and Heart Failure: Pharmacological Mechanisms. Expert Rev Clin Pharmacol. 2015;8(2):189–199. doi.org/10.1586/17512433.2015.1011125
Until the early 2000s, pharmaceutical companies were not obliged to publish all study results. Negative results often stayed in the drawer. Only since 2005 have leading journals required pre-registration of clinical trials. This created a bias whose full extent we are still working through. The first major statin studies of the 1990s and early 2000s fell exactly in this period. That does not mean their results are wrong. The effect of statins on heart attacks has been confirmed many times since. It only means we should read old studies in the context of their time.
And this belongs here too: if a cardiologist recommends statins to you, that happens on the basis of guidelines built from a great deal of evidence. In pure primary prevention these guidelines tie the recommendation to the calculated total risk, not to a single value. Ask about exactly that total risk in the conversation. It is the number that actually matters.
Do not ask your physician: “How much does this pill lower my LDL?” Ask: “How much does this pill lower my absolute risk of having a heart attack in the next ten years?” That is a different number. A much smaller number. It is the number that really matters.
What a modern cardiovascular risk profile looks like.
When someone comes to me with the question “am I cardiovascularly at risk?”, I do not look at total cholesterol alone. I look at the whole profile. I take a lot of time for this topic and I like to look at more than one lab value. Whether that view changes anything in your case can only be said once your values are on the table.
What can be checked in a differentiated cholesterol diagnostic
- Apo-B as a measure of the number of atherogenic particles
- Lp(a) at least once in a lifetime, because it is genetic
- LDL including subclasses, that is phenotype A or B (SD-LDL, MD-LDL, LD-LDL)
- HDL and HDL subclasses, because high HDL values allow more LDL before risk rises
- Triglycerides and the triglyceride-to-HDL ratio as insulin resistance markers
- Fasting insulin and HOMA-IR
- HbA1c
- hsCRP as an unspecific inflammation marker
- Homocysteine, vitamin B12, folate, and vitamin D
- Waist circumference, blood pressure, sleep quality, HRV
- oxLDL and Omega-3 Index
- If suspected: coronary calcium score (CAC) as imaging risk assessment. This involves radiation exposure, small though it is. The indication belongs in a medical assessment first.
That sounds elaborate. It is elaborate. But it shows more than a single number. One value alone is rarely enough.
Honesty includes the money: most of these values, that is Apo-B, Lp(a), LDL subclasses, oxLDL, omega-3 index, and the coronary calcium score, are not covered by German statutory health insurance. They are self-pay services. You get written information and a cost estimate beforehand and decide at your leisure. And once more: extended diagnostics do not replace a therapy that is indicated by guideline. They complement it.
You do not have to choose between eggs and your heart. That choice was never cleanly put to you.
If this article should accomplish one single thing, it is this: to free you from the wrong question.
The wrong question runs “Do I take butter or margarine today? Eggs or oats?”
The right question runs: “What pulls my endothelium out of balance every day? Where are the real stressors? Which can I influence?”
Energy, clarity, years of life. Those are the real values. Not a lowered LDL value on a lab slip.
This freedom is what I found for myself. My energy is balanced, my sleep quality is good. I eat meat, I eat eggs, I avoid refined seed oils, I take care of my insulin sensitivity. And I regularly measure the values that interest me. Apo-B. Lp(a). Triglycerides. Fasting insulin. HbA1c. That is my way. It does not have to be yours.
Three things you can do this week.
First, take a look at your oils. Look in your cupboard and at the ingredient list of your favorite snacks to see how much refined sunflower, safflower, soy, and corn oil is in there. You will be amazed where linoleic acid hides. Extra virgin olive oil or canola oil can be good alternatives here. Canola, incidentally, is low in linoleic acid and rich in alpha-linolenic acid, it does not belong in the same drawer as sunflower oil.
With butter, ghee, tallow, and coconut oil I am more cautious: in some people they can raise LDL considerably. If you have a familial lipid disorder or already have heart disease, please discuss this point with a physician first.
Second, talk about the further values at your next appointment. Apo-B. Lp(a) once. Triglycerides. Fasting insulin. HbA1c. Many practices are happy to run them when you bring it up. Part of this is not covered by insurance but is a self-pay service. You should know beforehand what it costs.
Third, look at your insulin sensitivity. Fewer fast carbohydrates, especially in the evening. Movement into daily life, ideally before or right after meals. Regular sleep in the dark hours. Stress hygiene in the evening. These four levers can favourably influence your metabolic profile.
Important if you take medication: anyone on insulin, sulfonylureas, or SGLT2 inhibitors must not sharply reduce carbohydrates without medical supervision. Otherwise there is a risk of hypoglycemia or ketoacidosis. Discuss this with your practice beforehand. The same applies with kidney or liver disease if you want to eat considerably more protein.
If you do not just want to read but really understand where your own cardiovascular risk lies, you can book an appointment below this article. We will go through your values together. With the differentiated view this topic deserves.
Sources
Re-analyses of historical RCTs
Ramsden CE, Zamora D, Leelarthaepin B, et al. Use of dietary linoleic acid for secondary prevention of coronary heart disease and death. BMJ. 2013;346:e8707. doi.org/10.1136/bmj.e8707 [RCT reanalysis, n=458, secondary prevention]
Ramsden CE, Zamora D, Majchrzak-Hong S, et al. Re-evaluation of the traditional diet-heart hypothesis: Minnesota Coronary Experiment (1968–73). BMJ. 2016;353:i1246. doi.org/10.1136/bmj.i1246 [RCT reanalysis + meta-analysis, n=9,423]
LDL subclasses, Apo-B, Lp(a)
Castelli WP. Lipids, risk factors and ischaemic heart disease. Atherosclerosis. 1996;124 Suppl:S1–9. doi.org/10.1016/0021-9150(96)05851-0 [Framingham overview]
Vekic J, Zeljkovic A, Cicero AFG, et al. Atherosclerosis Development and Progression: The Role of Atherogenic Small, Dense LDL. Medicina (Kaunas). 2022;58(2):299. doi.org/10.3390/medicina58020299 [Review]
Vekic J, Zeljkovic A, Al Rasadi K, et al. A New Look at Novel Cardiovascular Risk Biomarkers. Metabolites. 2022;12(2):108. doi.org/10.3390/metabo12020108 [Review]
Sniderman A, Kwiterovich PO. Update on the detection and treatment of atherogenic low-density lipoproteins. Curr Opin Endocrinol Diabetes Obes. 2013;20(2):140–147. doi.org/10.1097/MED.0b013e32835ed9cb [Review, Apo-B vs cholesterol markers]
Kamstrup PR. Lipoprotein(a) and Cardiovascular Disease. Clin Chem. 2021;67(1):154–166. doi.org/10.1093/clinchem/hvaa247 [Review, Mendelian randomization]
Griffin BA. Lipoprotein Atherogenicity: An Overview of Current Mechanisms. Proc Nutr Soc. 1999;58(1):163–169. doi.org/10.1079/pns19990022 [Review]
Hirano T. Pathophysiology of Diabetic Dyslipidemia. J Atheroscler Thromb. 2018;25(9):771–782. doi.org/10.5551/jat.RV17023 [Review]
Consensus and guidelines
Ference BA, Ginsberg HN, Graham I, et al. Low-density lipoproteins cause atherosclerotic cardiovascular disease. EAS Consensus. Eur Heart J. 2017;38(32):2459–2472. doi.org/10.1093/eurheartj/ehx144 [EAS consensus, >2 million participants]
Mach F, Baigent C, Catapano AL, et al. 2019 ESC/EAS Guidelines for the management of dyslipidaemias. Eur Heart J. 2020;41(1):111–188. doi.org/10.1093/eurheartj/ehz455 [Guideline]
Counter-position to LDL as universal risk
Ravnskov U, Diamond DM, Hama R, et al. Lack of an association or an inverse association between LDL-C and mortality in the elderly. BMJ Open. 2016;6(6):e010401. doi.org/10.1136/bmjopen-2015-010401 [Systematic review, n=68,094]
Soto-Mota A, Norwitz NG, Manubolu VS, et al. KETO-CTA: Plaque Predicts Plaque, ApoB Does Not. JACC Adv. 2025;4(7):101686. RETRACTED. Retraction notice: JACC Adv. 2026;5(5):102824. doi.org/10.1016/j.jacadv.2026.102824 [No longer used as evidence in this article]
Norwitz NG, Mindrum MR, Giral P, et al. Elevated LDL-cholesterol levels among lean mass hyper-responders. J Clin Lipidol. 2022;16(6):765–768. doi.org/10.1016/j.jacl.2022.10.010 [Editorial, LMHR]
Janič M, Lunder M, Janež A, et al. Rosuvastatin Improved LDL Subfractions Profile in a Patient with Type 1 Diabetes Following a Ketogenic Diet. Curr Vasc Pharmacol. 2025. doi.org/10.2174/0115701611370578250621183337 [Case report]
Fatty acids and endothelium
DiNicolantonio JJ, O’Keefe JH. Omega-6 vegetable oils as a driver of coronary heart disease: the oxidized linoleic acid hypothesis. Open Heart. 2018;5(2):e000898. doi.org/10.1136/openhrt-2018-000898 [Mechanism review]
Reaven P, Parthasarathy S, Grasse BJ, et al. Effects of Oleate-Rich and Linoleate-Rich Diets on the Susceptibility of Low Density Lipoprotein to Oxidative Modification. J Clin Invest. 1993;91(2):668–676. doi.org/10.1172/JCI116247 [Human study]
Berry EM, Eisenberg S, Haratz D, et al. Jerusalem Nutrition Study. Am J Clin Nutr. 1991;53(4):899–907. doi.org/10.1093/ajcn/53.4.899 [RCT, crossover]
Afonso MS, Castilho G, Lavrador MS, et al. The impact of dietary fatty acids on macrophage cholesterol homeostasis. J Nutr Biochem. 2014;25(2):95–103. doi.org/10.1016/j.jnutbio.2013.10.001 [Review, mechanism]
Statins
Cai T, Abel L, Langford O, et al. Associations between statins and adverse events in primary prevention. BMJ. 2021;374:n1537. doi.org/10.1136/bmj.n1537 [Meta-analysis, 62 RCTs, n=120,456]
Sattar N, Preiss D, Murray HM, et al. Statins and risk of incident diabetes. Lancet. 2010;375(9716):735–742. doi.org/10.1016/S0140-6736(09)61965-6 [Meta-analysis, 13 RCTs, n=91,140]
Okuyama H, Langsjoen PH, Hamazaki T, et al. Statins Stimulate Atherosclerosis and Heart Failure. Expert Rev Clin Pharmacol. 2015;8(2):189–199. doi.org/10.1586/17512433.2015.1011125 [Mechanism review]
Nutritional factors
Johnston BC, Zeraatkar D, Han MA, et al. Unprocessed Red Meat and Processed Meat Consumption: Dietary Guideline Recommendations From the Nutritional Recommendations (NutriRECS) Consortium. Ann Intern Med. 2019;171(10):756–764. doi.org/10.7326/M19-1621 [Guideline recommendation based on 5 systematic reviews]
Dinu M, Abbate R, Gensini GF, Casini A, Sofi F. Vegetarian, vegan diets and multiple health outcomes. Crit Rev Food Sci Nutr. 2017;57(17):3640–3649. doi.org/10.1080/10408398.2016.1138447 [Meta-analysis, k=96 studies]
Zhao B, Gan L, Graubard BI, et al. Dietary Cholesterol, Serum Cholesterol, and Egg Consumption With Mortality. Circulation. 2022;145(20):1506–1520. doi.org/10.1161/CIRCULATIONAHA.121.057642 [Meta-analysis, n=3,601,401, counter-position]
Important for your safety. This article does not replace a medical examination, a diagnosis, or a treatment. Never stop prescribed medication on your own. With acute symptoms such as chest pain, pressure on the chest, or shortness of breath, call the emergency number immediately (112 in Europe, 911 in the US).
Editorial note. This text has been revised. One originally cited paper (KETO-CTA, JACC Advances 2025) was retracted after publication at the request of the authors and the editors and has therefore been removed. Several study summaries were made more precise so that they reproduce the conclusions of the original papers in full.