Ketogenic Diet and High LDL Cholesterol: Is It Dangerous?

The first and most obvious explanation concerns the composition of the diet. Keto usually means a higher intake of saturated fat (butter, fatty meat, cream, coconut oil), and saturated fatty acids have been known for decades to raise LDL by lowering the activity of LDL receptors in the liver. Added to this is a higher intake of dietary cholesterol (eggs, offal), which affects blood levels more strongly in some people, the so-called cholesterol hyper-responders. In people eating in a calorie deficit and losing weight quickly, mobilization of body fat also temporarily raises blood cholesterol regardless of what the diet contains.

The second explanation, popular in the biohacker community, is the 'lipid energy model'. It assumes that in lean people, who have small fat stores, the liver on a ketogenic diet increases the transport of lipids (VLDL and, consequently, LDL) to peripheral tissues as an energy source, so a rising LDL would reflect faster fuel turnover rather than a problem. This is a mechanistic hypothesis. It is sometimes presented as established fact, but no data show that LDL particles in this setting are less atherogenic. We come back to what this means in practice when we discuss the LMHR phenotype.

The third layer is individual variation. Some people on keto have no rise in LDL at all or only a small one, while in others LDL increases by 100 to 300 mg/dL. Age, body weight, genetics (variants of LDLR, APOE and PCSK9), thyroid function and the amount of saturated fat all contribute to this spread. That is why a single average from studies says little about a specific person, and why your own repeated measurement matters so much.

Two limitations must be stated honestly. First, the number of studies in the meta-analysis was very small (three) and they were short, so this is a signal of direction, not a precise measure. Second, these are average values: behind a mean of +42 mg/dL are people whose LDL barely changed and people whose LDL rose by several hundred mg/dL. Broader reviews put the range of mean LDL-C changes in short trials in people with a normal BMI at roughly 18-70 mg/dL.

The rise in ApoB matters too, because it, rather than LDL-C itself, counts atherogenic particles. The meta-analysis suggests that, at least in these three trials, it was not just an 'artifact of measuring cholesterol in the particle': the number of ApoB-carrying particles also increased. So the argument that 'LDL-C rises but the number of particles stays the same' finds no confirmation in the data.

The term LMHR describes lean, metabolically healthy people with high HDL and low triglycerides whose LDL rises especially sharply on a ketogenic diet. In studies of this group the typical definition is LDL-C of 190 mg/dL (4.9 mmol/L) or more, HDL-C of 60 mg/dL or more and triglycerides of 80 mg/dL or less. It is not a medical diagnosis or a disease entity recognized in guidelines. It is a working description created mainly within the low-carbohydrate community and a narrow group of researchers closely associated with it.

What follows from this? That the rise in LDL in lean people is a real, repeatable physiological phenomenon that depends on body composition and metabolism, not only on the amount of saturated fat. It does not follow, however, that it is safe. The mechanism by which a high LDL arises says nothing about whether the LDL is dangerous: an ApoB-carrying particle that enters the arterial wall does so regardless of why it ended up in the blood. The hypothesis of a 'harmless' LDL therefore rests on a logical leap, from explaining the cause to a conclusion about the consequence.

There are also contrary reports. A case series published in 2026 from a German lipid center described three patients referred with suspected familial hypercholesterolemia, whose LDL-C on a keto diet reached between 364 and 682 mg/dL. None consistently met all the LMHR criteria (one was overweight before the diet and one became clearly underweight), and after a moderate increase in carbohydrates their LDL-C and ApoB fell markedly without medication. The authors stress that severe diet-induced hypercholesterolemia can mimic familial hypercholesterolemia and that a high ApoB should not be assumed to be benign.

This study is sometimes cited as proof that a high LDL on a ketogenic diet is harmless. That reading goes beyond what the study can show. It is cross-sectional: it assesses the arteries at a single point in time, with no imaging from before the diet, so we do not know whether plaque before keto was smaller, the same or larger. Participants came forward on their own (selection: those who tolerated a high LDL poorly or who had plaque may have changed their diet or not joined at all), and a mean exposure of 4.7 years is a fraction of the period over which atherosclerosis develops across a lifetime. In addition, both groups had a relatively low average plaque burden, and the study did not measure heart attacks or deaths.

The key word in this consensus is 'cumulative': risk is determined not only by the LDL value but by the total time it persists. A person who spends 10 years at an LDL of 250 mg/dL accumulates a much larger 'dose' of LDL than a person with an LDL of 100 mg/dL over the same period, even though the arteries of both may look similar today. No plaque after a few years is therefore information about the past, not a guarantee for the coming decades. On the other hand, no one has proven that LDL produced by keto is an exception to this rule; there simply are no data that address it.

This is summed up well by a review published in 2026 in the Journal of the American Heart Association: evidence that diet-induced LDL-C elevation in people with a normal BMI is harmless does not exist; the limited imaging data do not establish cardiovascular safety; and the long-term atherosclerotic consequences in this population have not been studied directly. The authors recommend regular lipid monitoring and an individual conversation about risk, especially in lean people who appear low-risk at first glance.

LDL-C on its own can be misleading on a high-fat diet: the Friedewald formula assumes a particular particle composition and can distort the result, especially with very low triglycerides and highly variable particles. The most informative supplement is apolipoprotein B (ApoB), which reflects the number of atherogenic particles. We describe it in detail in a separate knowledge-base entry and in an article comparing LDL and ApoB.

Two practical notes. First, if the result after starting keto is surprising, repeat the test after 8-12 weeks on a stable diet and stable body weight: during rapid weight loss the lipid panel can be temporarily distorted. Second, a single LDL above 190 mg/dL (4.9 mmol/L) should not be attributed to diet alone without excluding familial hypercholesterolemia; in that case the diet may only mask or amplify a genetic condition.

A meta-analysis of 60 controlled feeding trials showed that fatty acids affect lipids depending on type and amount, and that replacing saturated (and trans) fats with cis-unsaturated fats most effectively improves the ratio of total cholesterol to HDL. The authors also noted that effects on lipid markers do not automatically translate into risk and need confirmation in trials with clinical endpoints. This is important context: not every fat in a ketogenic diet acts the same way.

In practice, 'keto' is not one diet. A version built on olive oil, avocado, nuts and fatty sea fish has a completely different fatty acid profile from one built on butter, fatty red meat, cream and coconut oil. The study in lean women suggested that, among the factors predicting the change in LDL, saturated fat did not stand out from the others, but that is a small sample and does not excuse caution: for most people, if the rise in LDL is a problem, swapping saturated fat for monounsaturated and polyunsaturated fat is the first and cheapest step.

This is one of three common interpretation traps. The second is the belief that 'if my CAC is 0, I can ignore LDL': CAC measures calcium, the later, calcified stage of plaque, and in younger people a score of 0 can be normal despite growing risk. The third is equating no symptoms with no disease: atherosclerosis develops silently for years, and the first symptom is sometimes a heart attack. We write more about the role of HDL in the article on low HDL and in knowledge-base entries on the lipid panel.

There is no official guideline LDL value at which you should 'end keto'. Dyslipidemia guidelines do not address this phenotype directly; they treat LDL and ApoB according to cardiovascular risk profile regardless of cause. In practice the decision is an honest weighing of benefits of the diet (weight loss, glycemic control) against the burden on the arteries from a high LDL and ApoB over the years. In a person with type 2 diabetes or insulin resistance, the benefits of keto may outweigh the rise in LDL; in a lean, healthy person following it 'for optimization', the balance is less clear.

A ketogenic diet has legitimate uses and for some people is a very effective tool. But a high LDL on keto is neither a verdict nor a harmless curiosity. The most sensible stance is to treat it like any other high LDL: measure ApoB and Lp(a), exclude secondary causes and familial hypercholesterolemia, assess overall risk and, if the rise is large, adjust the composition of the diet rather than assume that the physiology of a lean mass hyper-responder 'overrides' the biology of LDL. Absence of evidence of harm is not evidence of absence of harm, especially when the mechanism is so well known.