Cholesterol

Lipoproteins are particles with a fatty core, containing cholesterol and triglycerides, wrapped in a shell of proteins called apolipoproteins. LDL carries most of the cholesterol in the blood to the tissues; very-low-density lipoprotein (VLDL) carries triglycerides made in the liver; high-density lipoprotein (HDL) carries cholesterol away from tissues. LDL, VLDL, their remnants and lipoprotein(a) each carry one molecule of apolipoprotein B (apoB), so the apoB concentration counts the number of these particles. Cholesterol enters the artery wall only inside apoB particles, and the number of particles in the blood largely determines how many become trapped in the wall; apart from lipoprotein(a) and certain abnormal remnants, all apoB particles are considered equally atherogenic (plaque-forming) [8].

A standard lipid panel reports total cholesterol, HDL cholesterol and triglycerides, and calculates or measures LDL cholesterol. Non-HDL cholesterol, total minus HDL cholesterol, captures the cholesterol in all apoB particles. Results are given in mmol/L or mg/dL; 1 mmol/L of LDL cholesterol equals about 38.7 mg/dL [3].

In 346,686 UK Biobank participants not taking statins, followed for a median of 8.9 years, apoB, directly measured LDL cholesterol and non-HDL cholesterol were highly correlated and had similar associations with cardiovascular events (hazard ratios of 1.23, 1.20 and 1.21 per standard deviation higher). Once total and HDL cholesterol were included in a risk model, adding apoB or LDL cholesterol gave no substantive improvement in prediction [9]. Proponents of apoB testing argue that, because the amount of cholesterol per particle varies, a particle count measures atherogenic risk more accurately than a cholesterol mass [8].

Several kinds of evidence point to LDL as a cause of atherosclerosis rather than a marker of it. Rare mutations that reduce LDL receptor function produce very high LDL cholesterol and early heart disease, while rare variants that lower LDL cholesterol protect against it. Across prospective studies, Mendelian randomisation studies (which compare people who inherited different gene variants, a natural experiment) and randomised trials, the relationship between LDL cholesterol and risk is log-linear (each fixed reduction lowers risk by a similar proportion) and grows stronger with longer exposure [1].

A Mendelian randomisation analysis of nine gene variants in six genes, combining data from 312,321 participants, found 54.5% lower risk of coronary heart disease for each 1 mmol/L lower LDL cholesterol, with no difference between variants, three times the reduction per unit seen in statin trials [3]. Among 438,952 UK Biobank participants, gene variants giving 14.7 mg/dL lower LDL cholesterol were associated with an odds ratio of 0.73 for major coronary events, and combined lifelong exposure to 38.67 mg/dL lower LDL cholesterol and 10 mm Hg lower systolic blood pressure was associated with an odds ratio of 0.22 [10].

Familial hypercholesterolaemia is an inherited condition in which LDL cholesterol is high from birth. Its heterozygous form was estimated at 1 in 500 people, and direct screening in a northern European population found about 1 in 200. Fewer than 1% of cases are diagnosed in most countries, and the risk of coronary heart disease is up to 13 times higher when it is untreated [11]. The broader disease picture is described under cardiovascular disease.

Statins block an enzyme the liver uses to make cholesterol, which increases the uptake of LDL from the blood. The Cholesterol Treatment Trialists' Collaboration analysed individual data from 21 trials of statin against control (129,526 participants) and five trials of more against less intensive statin treatment (39,612 participants). More intensive treatment lowered LDL cholesterol by a further 0.51 mmol/L and major vascular events by a further 15%. Across all trials, each 1.0 mmol/L reduction lowered major vascular events by 22% (rate ratio 0.78) and all-cause mortality by 10%, mainly through fewer deaths from coronary heart disease. There was no increase in cancer or non-vascular deaths and no threshold below which lowering stopped helping, and the authors estimated that a reduction of 2 to 3 mmol/L would lower risk by about 40 to 50% [2].

Drugs that lower LDL cholesterol by other routes give benefits in proportion to the reduction. In IMPROVE-IT, adding ezetimibe, which reduces cholesterol absorption from the gut, to a statin in 18,144 patients after an acute coronary syndrome lowered average LDL cholesterol to 53.7 mg/dL compared with 69.5 mg/dL, and the seven-year rate of cardiovascular events from 34.7% to 32.7% [12]. In FOURIER, the PCSK9 inhibitor evolocumab, an injected antibody, lowered LDL cholesterol by 59% from a median of 92 to 30 mg/dL in 27,564 patients with cardiovascular disease already taking statins, and reduced cardiovascular death, heart attack or stroke by 20% over a median of 2.2 years [5].

The type of dietary fat affects blood cholesterol. A pooled analysis of 60 controlled feeding trials found that replacing saturated fat with unsaturated fat lowered the ratio of total to HDL cholesterol, while replacing it with carbohydrate left the ratio unchanged and raised triglycerides [14]. A 2020 Cochrane review of 15 randomised trials with about 59,000 participants found that reducing saturated fat for at least two years lowered combined cardiovascular events by 21% (risk ratio 0.79), with greater reductions in trials that lowered cholesterol more, but had little or no effect on all-cause or cardiovascular mortality [15].

Soluble fibre binds bile acids in the gut, which the liver replaces by drawing cholesterol from the blood. A pooled analysis of 28 randomised trials found that at least 3 g a day of oat beta-glucan lowered LDL cholesterol by 0.25 mmol/L without changing HDL cholesterol or triglycerides [16]. Combining several cholesterol-lowering foods produces larger effects. In a one-month trial in 46 adults with high cholesterol, a "portfolio" diet high in plant sterols, soy protein, viscous fibre and almonds lowered LDL cholesterol by 28.6%, compared with 8.0% for a low-saturated-fat control diet and 30.9% for the same diet plus 20 mg of lovastatin; the difference between the portfolio diet and the statin was not significant [4]. Eating patterns are described under dietary patterns.

A 1992 pooled analysis of 70 studies of weight loss by dieting found significant decreases in total cholesterol, LDL cholesterol, VLDL cholesterol and triglycerides. HDL cholesterol fell slightly while weight was being lost and rose by about 0.009 mmol/L per kilogram lost once weight had stabilised [17]. See body weight.

Exercise changes the lipoprotein profile more than the LDL cholesterol level. The STRRIDE trial randomly assigned 111 sedentary, overweight adults with mildly abnormal lipids to a control group or to one of three exercise programmes for about eight months, while keeping body weight stable. The largest amount of exercise, equivalent to jogging about 32 km a week, improved 10 of 11 lipoprotein measures more than lower amounts did, all three programmes did better than the control group, and the improvements depended on the amount of exercise rather than its intensity or the gain in fitness [18]. The wider effects of activity are covered under exercise and longevity.

In observational studies, higher HDL cholesterol is associated with lower heart attack risk, but raising it has not been shown to reduce risk. A Mendelian randomisation study found that carriers of a gene variant raising HDL cholesterol by 0.14 mmol/L, a difference expected to lower heart attack risk by 13%, had no difference in risk (odds ratio 0.99), and a genetic score for higher HDL was also not associated with risk, whereas a genetic score for LDL cholesterol was [6]. In the AIM-HIGH trial of 3,414 patients with cardiovascular disease on statins, niacin raised median HDL cholesterol from 35 to 42 mg/dL and lowered triglycerides, but the trial was stopped after three years for lack of benefit, with events in 16.4% of the niacin group and 16.2% of the placebo group [13].

Lipoprotein(a), or Lp(a), is an LDL-like particle with an extra protein attached. More than 90% of the variation in its level between people is determined by genetic variation at the LPA gene; the level reaches adult values by about age five and is changed little by lifestyle [7]. In the Copenhagen City Heart Study, people with Lp(a) above the 95th percentile had 2.6 times the heart attack risk of those below the 22nd percentile, and gene variants that raise Lp(a) were associated with a 22% higher risk per doubling of its level, consistent with a causal effect [19]. A 2022 European Atherosclerosis Society consensus statement described the association as causal and continuous, estimated that a level of 100 mg/dL roughly doubles the risk of atherosclerotic cardiovascular disease, linked high Lp(a) with calcification of the aortic valve, and recommended measuring Lp(a) at least once in every adult. No specific Lp(a)-lowering drug had yet been shown to reduce events [7].

Mendelian randomisation depends on the assumption that the gene variants affect heart disease only through the lipid being studied. The authors of the UK Biobank analysis noted that lifelong genetic differences cannot be assumed to show the benefit achievable by treatment started in adulthood [10].

Diet trials of cholesterol are short and small compared with drug trials. The portfolio trial lasted one month in 46 people [4], the oat beta-glucan trials lasted 2 to 12 weeks [16], and the Cochrane review of saturated fat found a reduction in cardiovascular events but not in deaths [15]. Lipid changes measured in feeding trials are risk markers, and their authors cautioned that they should be confirmed by studies of disease outcomes [14].

The statin and PCSK9 trials enrolled mostly people with existing cardiovascular disease or high risk, and followed them for years rather than decades [2, 5]. The long-term effect of starting treatment in young adults with modestly raised LDL cholesterol has been estimated from genetic studies rather than tested directly.

Full deep dive: cholesterol.