Does LDL cholesterol actually cause heart disease?
Higher lifelong exposure to LDL / ApoB-bearing lipoproteins is an independent causal contributor to atherosclerotic cardiovascular disease (ASCVD).
Mainstream science treats LDL as a genuine cause of artery disease, supported by genetics, drug trials, and large population studies that point the same way. A vocal minority argues the causal claim is overstated, especially in older and metabolically healthy people. The honest reading: LDL is very likely a real causal contributor, but how much it matters for any one person depends heavily on their baseline risk and metabolic context.
Evidence balance
Mainstream steelman
The causal case rests on a triangulation rarely available in medicine. Mendelian randomization shows that genetic variants lowering LDL from birth lower lifetime ASCVD risk in a dose-dependent way proportional to cumulative exposure. Randomized trials across unrelated drug classes (statins, ezetimibe, PCSK9 inhibitors) reduce events roughly in proportion to the absolute LDL reduction achieved, regardless of mechanism - exactly what you expect if LDL itself is on the causal path. And consistent prospective epidemiology spanning more than 200 studies and millions of participants points the same way. When genetic, interventional, and observational evidence converge on the same dose-response, coincidence becomes very hard to argue.
Skeptic steelman
Skeptics argue the causal claim is overstated. In people over 60 - most of those who actually die of heart disease - higher LDL-C is repeatedly associated with equal or lower all-cause mortality. Randomized trials report benefit overwhelmingly in relative terms; absolute risk reductions in primary prevention are often small. Mendelian randomization has real limits: pleiotropy, weak instruments, mostly European-ancestry samples, and canalization - a lifelong genetic nudge is not the same as a drug started at 55. And the lean-mass-hyper-responder phenotype shows that very high LDL can coexist with excellent metabolic markers and, so far, little measurable plaque, suggesting LDL's danger is conditional, not automatic.
Bottom line
High confidenceThe triangulated evidence (genetics plus randomized trials plus epidemiology) makes LDL / ApoB-bearing particles a genuine causal contributor to atherosclerosis - this is hard to dismiss honestly. But "causal contributor" is not "sole or context-free cause": absolute risk depends heavily on baseline risk and metabolic context, and the elderly and LMHR anomalies are real and must be engaged, not waved away.
This is a clearly-labelled editorial judgment, not a fact. It is written to survive its own skeptic steelman above.
What would change this conclusion
A well-powered randomized trial (or strong natural experiment) showing that lowering LDL/ApoB in genuinely metabolically-healthy, otherwise-low-risk people produces no reduction in hard cardiovascular events; or robust evidence that the Mendelian-randomization dose-response is an artifact of pleiotropy.
The evidence (8)
Strongest evidence first. Each card traces to a study and a verbatim quote with a locator.
- SupportsHigh tierMajor adverse cardiovascular event
Efficacy and safety of more intensive lowering of LDL cholesterol: a meta-analysis of data from 170,000 participants in 26 randomised trials
The Lancet, 2010 - Meta Analysis Of Rcts
A large individual-participant meta-analysis showing that lowering LDL reduces major vascular events dose-dependently, with no threshold, supports the causal interpretation from the intervention side: change the exposure, change the outcome. It is supporting rather than decisive for causality because statins have effects beyond LDL, though concordance with non-statin lowering mitigates this.
“rate ratio [RR] 0·78, 95% CI 0·76-0·80; p<0·0001”
Applicability: Mixed primary and secondary prevention populations.
- A drug effect does not perfectly isolate LDL from statins' pleiotropic effects.
- Strongly supportsHigh tierComposite clinical outcome
Low-density lipoproteins cause atherosclerotic cardiovascular disease. 1. Evidence from genetic, epidemiologic, and clinical studies. A consensus statement from the European Atherosclerosis Society Consensus Panel
European Heart Journal, 2017 - Guideline Or Consensus
The EAS panel explicitly concludes that triangulated genetic, epidemiologic, and clinical evidence establishes LDL as a cause of ASCVD - the single strongest institutional statement of the causal claim, resting on a very large pooled evidence base.
“Consistent evidence from numerous and multiple different types of clinical and genetic studies unequivocally establishes that LDL causes ASCVD.”
Applicability: General adult populations; the statement does not address metabolically-healthy very-high-LDL phenotypes such as LMHR.
- A consensus statement (advocacy of a position), not itself a new experiment.
- Several panel members have pharmaceutical relationships.
- Strongly supportsHigh tierCoronary event
Effect of long-term exposure to lower low-density lipoprotein cholesterol beginning early in life on the risk of coronary heart disease: a Mendelian randomization analysis
Journal of the American College of Cardiology, 2012 - Mendelian Randomization
Mendelian randomization uses randomly-allocated genotypes to estimate the effect of lifelong lower LDL, sharply reducing confounding and reverse causation. The consistent dose-response across nine independent variants is strong evidence that LDL is on the causal path, not merely correlated.
“Prolonged exposure to lower LDL-C beginning early in life is associated with a substantially greater reduction in the risk of CHD than the current practice of lowering LDL-C beginning later in life.”
Applicability: Predominantly European-ancestry genetic samples; lifelong genetic exposure (canalization) differs from drug therapy started in mid-life.
- MR assumes no pleiotropy and adequate instrument strength.
- SupportsHigh tierComposite clinical outcome
Evolocumab and Clinical Outcomes in Patients with Cardiovascular Disease
New England Journal of Medicine, 2017 - Randomized Controlled Trial
FOURIER strengthens the causal case from yet another angle: PCSK9 inhibition - a mechanism distinct from both statins and ezetimibe - significantly reduced a hard clinical composite, with no safety floor at very low LDL. Three unrelated mechanisms all lowering LDL and all reducing events is the convergent pattern expected if LDL is causal. The short follow-up leaves the mortality question open, but the direction of the event effect is clear.
“Relative to placebo, evolocumab treatment significantly reduced the risk of the primary end point (1344 patients [9.8%] vs. 1563 patients [11.3%]; hazard ratio, 0.85; 95% confidence interval [CI], 0.79 to 0.92; P<0.001) and the key secondary end point (816 [5.9%] vs. 1013 [7.4%]; hazard ratio, 0.80; 95% CI, 0.73 to 0.88; P<0.001).”
Applicability: Secondary prevention; the causal inference rests on the clinical event reduction via a third distinct mechanism.
- Short median follow-up (2.2 years), underpowered for mortality; manufacturer-funded (Amgen).
- SupportsHigh tierComposite clinical outcome
Ezetimibe Added to Statin Therapy after Acute Coronary Syndromes
New England Journal of Medicine, 2015 - Randomized Controlled Trial
IMPROVE-IT adds intervention-side support for LDL causality through a non-statin route. Because the only thing ezetimibe meaningfully does is lower LDL absorption, its significant reduction of a hard clinical composite points to LDL itself as the operative exposure rather than a pleiotropic statin effect. The causal weight rests on this clinical event reduction, not on the LDL number alone.
“The Kaplan-Meier event rate for the primary end point at 7 years was 32.7% in the simvastatin-ezetimibe group, as compared with 34.7% in the simvastatin-monotherapy group (absolute risk difference, 2.0 percentage points; hazard ratio, 0.936; 95% confidence interval, 0.89 to 0.99; P=0.016).”
Applicability: Secondary prevention (post-ACS); the causal inference rests on the clinical event reduction via a non-statin mechanism.
- Small absolute effect; upper confidence bound near 1.0; manufacturer-funded (Merck).
- Strongly supportsHigh tierCoronary event
Diagnostic Yield and Clinical Utility of Sequencing Familial Hypercholesterolemia Genes in Patients With Severe Hypercholesterolemia
Journal of the American College of Cardiology, 2016 - Case Control
A natural-experiment design that isolates lifelong LDL exposure. At the same measured LDL (>=190 mg/dl), carrying an FH mutation - which raises LDL from birth - conferred a 22-fold CAD risk versus 6-fold without the mutation. Same current LDL, very different cumulative exposure, very different risk: that is direct support for the cumulative-causal-exposure model and strongly supports LDL causality.
“Compared with a reference group with LDL cholesterol <130 mg/dl and no mutation, participants with LDL cholesterol >=190 mg/dl and no FH mutation had a 6-fold higher risk for CAD (odds ratio: 6.0; 95% confidence interval: 5.2 to 6.9), whereas those with both LDL cholesterol >=190 mg/dl and an FH mutation demonstrated a 22-fold increased risk (odds ratio: 22.3; 95% confidence interval: 10.7 to 53.2).”
Applicability: Case-control with genetic stratification; speaks to cumulative LDL exposure, directly relevant to the causal claim.
- Case-control design; the FH-mutation contrast is the key causal lever rather than the absolute odds.
- ChallengesLow tierAll-cause mortality
Lack of an association or an inverse association between low-density-lipoprotein cholesterol and mortality in the elderly: a systematic review
BMJ Open, 2016 - Systematic Review
This is the strongest single piece of the "LDL skeptic" case applied to older people, and it challenges a universal causal reading: in these elderly cohorts no group showed the higher mortality with higher LDL that the hypothesis predicts, and most showed the opposite. It is entered at low tier and flagged because it is a skeptic-authored, Western Vascular Institute-funded vote-counting review with no pooled estimate that does not address reverse causation - serious illness lowers LDL and raises mortality - which can by itself manufacture an inverse association in the old.
“Inverse association between all-cause mortality and LDL-C was seen in 16 cohorts (in 14 with statistical significance) representing 92% of the number of participants, where this association was recorded. In the rest, no association was found.”
Applicability: Community-dwelling people aged 60+ and all-cause mortality - a different endpoint and population from the ASCVD-causality evidence; does not overturn the RCT/MR causal base.
- Skeptic-authored review (several authors wrote books criticising the cholesterol hypothesis); funded by the Western Vascular Institute.
- Crude vote-counting with no pooled effect estimate; heavily criticised post-publication (CEBM Oxford, UK Science Media Centre).
- Does not address reverse causation, which can produce an apparent inverse LDL-mortality association in the elderly.
- All-cause mortality in the elderly is a distinct question from whether LDL is causal for atherosclerotic disease.
- ChallengesVery low tierPlaque imaging (surrogate)
Carbohydrate Restriction-Induced Elevations in LDL-Cholesterol and Atherosclerosis: The KETO Trial
JACC: Advances, 2024 - Imaging Study
Within a metabolically-healthy very-high-LDL cohort, LDL-C did not correlate with coronary plaque, which is in tension with a simple, context-free dose- response. It only weakly challenges the general causal claim: it is cross- sectional, small, phenotype-specific, and measures plaque rather than events.
“Coronary plaque in metabolically healthy individuals with carbohydrate restriction-induced LDL-C ≥190 mg/dL on KETO for a mean of 4.7 years is not greater than a matched cohort with 149 mg/dL lower average LDL-C. There is no association between LDL-C and plaque burden in either cohort.”
Applicability: Only the LMHR/metabolically-healthy phenotype; cross-sectional baseline; surrogate (plaque) endpoint, not events.
Tier adjusted: Downgraded from the study's source tier (low) because, for the GENERAL causal claim, it is indirect: a small, cross-sectional, single-phenotype sample using a surrogate (plaque) rather than events.
- The one-year longitudinal follow-up was retracted (methodology concerns); no reliable longitudinal data are available.