Keto and Cholesterol: Reading the Lipid Evidence
A ketogenic diet can improve some lipid markers and worsen others. Across randomized trials, triglycerides fall and HDL cholesterol rises on average, while LDL cholesterol and total cholesterol rise on average. Individual LDL responses range widely and can be much larger than the pooled mean. A lower triglyceride result, higher HDL, weight loss, or measurable ketones does not cancel an LDL or apoB increase. Lipid results need interpretation alongside baseline values, family history, diabetes, blood pressure, smoking, kidney disease, known cardiovascular disease, medicines, and other risk information.
This page was written from current randomized trials, systematic evidence, and official cardiovascular guidance rather than biased interpretations of cholesterol research. It explains population evidence and questions for clinical review. It does not interpret your results, set a personal target, select a testing schedule, or advise starting, stopping, or changing medication.
“Cholesterol” is several measurements, not one score
A standard lipid panel reports measurements that answer different questions:
| Measurement | What it reflects | Why a keto change can be misread |
|---|---|---|
| LDL cholesterol | Cholesterol carried within LDL particles | A rise can be dismissed when triglycerides fall, even though the measures are not substitutes |
| Non-HDL cholesterol | Cholesterol in all apoB-containing particles combined | It can add context when several atherogenic particle classes are present |
| Triglycerides | A circulating form of fat influenced by recent intake, metabolism, diabetes control, alcohol, and other factors | A fall is favorable but does not prove LDL-related risk disappeared |
| HDL cholesterol | Cholesterol carried in HDL particles | A higher value is not a treatment that neutralizes elevated LDL or apoB |
| ApoB | The main structural protein present as one copy on most atherogenic particles | It estimates particle number and may clarify risk in selected clinical contexts |
The 2026 ACC/AHA dyslipidemia guideline keeps LDL cholesterol and non-HDL cholesterol central to risk-based management. It says apoB can be useful in selected situations, particularly when triglycerides are elevated, diabetes is present, or LDL has already reached a low value. The guideline also places any number inside a broader cardiovascular-risk assessment rather than treating diet as the only variable.4
That does not mean everyone needs every advanced test. It means a home ketone meter, body weight, or triglyceride-to-HDL ratio cannot stand in for a clinical lipid assessment.
What the newest randomized-trial meta-analysis found
Juanjuan Zhao, Chia Wei, Fao Gong, Zongran Pang, and Huanhu Zhao published a 2026 systematic review and meta-regression of 53 adult randomized controlled trials. Their ketogenic-diet definition required more than 45% of energy from fat and less than 10% from carbohydrate. The search ran through June 2025 and included varied populations, durations, and comparison diets.1
Compared with control diets, the pooled weighted mean differences were:
| Marker | Average difference with keto | Evidence certainty reported by the review |
|---|---|---|
| Triglycerides | 22.31 mg/dL lower | Moderate |
| HDL cholesterol | 3.52 mg/dL higher | Moderate |
| LDL cholesterol | 8.22 mg/dL higher | Low |
| Total cholesterol | 8.06 mg/dL higher | Low |
These are group averages, not a forecast. Heterogeneity was substantial, and prediction intervals were wide. For triglycerides, the 95% prediction interval extended from a large decrease to a small increase, meaning even the direction may differ in a future comparable study. The review also lacked long-term cardiovascular-event evidence. It can describe average marker changes under the included interventions; it cannot prove the net effect on heart attacks or strokes or tell one reader whether a diet is safe.1
Low certainty for LDL does not mean LDL is unimportant. It means confidence in that pooled size is limited by the underlying trials and inconsistency. The direction of the average result, the uncertainty around its magnitude, and the clinical importance of a person's measured response are three separate issues.
Why an average can hide a large response
Jonas Burén, Madelene Ericsson, Nágila Raquel Teixeira Damasceno, and Anna Sjödin conducted a controlled crossover feeding trial in healthy, young, normal-weight women. Twenty-four were assigned and 17 completed both four-week diet phases. The ketogenic low-carbohydrate, high-fat phase increased LDL cholesterol in every completer. ApoB, small dense LDL, and large buoyant LDL also increased.2
The trial was short, small, limited to women fitting a narrow profile, and used a specific high-fat menu. It cannot supply a universal effect size or establish long-term clinical outcomes. Its contribution is different: it demonstrates that a pronounced, consistent LDL and apoB response can occur under controlled feeding, even among young adults without overweight. Assuming that only older adults or people with obesity need monitoring would miss that finding.
At the other end, some people see modest or little LDL change. Weight change, baseline metabolic health, genetics, energy balance, saturated-fat intake, fiber, and the comparator diet may contribute to variation. Current research cannot turn those factors into a reliable online predictor for one person.
The Keto-Med trial shows the marker tradeoff directly
The keto versus Mediterranean diet comparison covers the full Keto-Med trial. For lipids, its main contribution is a direct 12-week crossover comparison in adults with prediabetes or type 2 diabetes. Triglycerides changed by minus 16% during the well-formulated ketogenic phase and minus 5% during the Mediterranean-plus phase. LDL cholesterol changed by plus 10% during keto and minus 5% during Mediterranean-plus. Both between-diet differences were statistically significant.3
The two phases shared non-starchy vegetables and excluded added sugar and refined grains. Mediterranean-plus also included legumes, fruit, and intact whole grains, while keto excluded them and delivered less fiber. This is more informative than comparing two labels in the abstract, but it remains 33 complete cases, 12 weeks per diet, and no cardiovascular events.
Food quality can change the pattern without guaranteeing the result
The American Heart Association's 2026 dietary guidance emphasizes an overall pattern rich in varied vegetables and fruits, whole grains, healthy protein sources, minimally processed foods, and fiber, with unsaturated-fat sources replacing saturated-fat sources.5 A strict ketogenic diet cannot reproduce every part of that pattern, but the fat-source and minimally processed-food principles can still guide a food-quality review.
Possible substitutions within a carbohydrate-restricted pattern include:
- olive, canola, avocado, or other non-tropical liquid plant oils in place of some butter, coconut oil, or cream;
- nuts, seeds, avocado, olives, and fish in place of some fatty processed meat;
- varied non-starchy vegetables and compatible intact plant foods instead of relying on refined low-carbohydrate snacks;
- the keto fiber guide to review intact foods and added fibers without turning fiber into an automatic supplement dose.
These changes bring the pattern closer to current cardiovascular dietary guidance. They do not guarantee that LDL or apoB will fall, and they are not a substitute for measuring what happened.
How to prepare for a useful clinical conversation
Bring the actual results and context, not a conclusion built from one number. Useful information can include:
- The baseline lipid panel and the result obtained after the dietary change, with dates and whether the same laboratory and preparation instructions were used.
- The main foods eaten before and during keto, especially fat sources, plant foods, packaged products, alcohol, and major weight change.
- Current medicines and supplements. Do not stop a statin or another lipid-lowering medicine to “test” the diet unless the prescriber directs it.
- Personal history of cardiovascular disease, diabetes, hypertension, kidney disease, pregnancy complications, or smoking.
- Family history of very high cholesterol or premature heart attack or stroke, which may raise concern for familial or inherited risk.
- Questions about whether the standard panel is sufficient, whether selective apoB or lipoprotein(a) testing would add information, and when the clinician wants results repeated.
This page cannot determine whether a value is an emergency, whether a diet should continue, or whether medication is indicated. Markedly abnormal or unexpected results, chest pain, shortness of breath, fainting, or signs of stroke need timely professional care, not diet experimentation.
Claims the evidence does not support
- “HDL went up, so LDL no longer matters.” The particles and measurements have different roles.
- “Triglycerides fell, so cardiovascular risk must be lower overall.” The trials mostly measured markers, not clinical events.
- “Large LDL particles are harmless.” The Burén trial found apoB and both large and small LDL fractions rose; particle size does not erase particle number.
- “Only saturated fat matters.” Fat source is important, but genetics, energy balance, weight change, fiber, health conditions, and the full dietary pattern can also influence results.
- “Everyone on keto needs medication” or “no one on keto needs medication.” Treatment decisions belong to a risk-based clinician-patient discussion under current guidelines.
Frequently Asked Questions
Does keto always raise LDL cholesterol?
Do lower triglycerides cancel a higher LDL result?
Should everyone on keto get an apoB test?
Should I stop my statin if I change my diet?
Works cited
Article history
- First published
Your learning guide
Loading reading status…