Cholesterol and lipoproteins
Heart disease is the biggest single barrier to a long life, and the cholesterol behind it is widely misunderstood. What ages your arteries is not the number on one panel but how many cholesterol-carrying particles your blood vessels have been exposed to, and for how many years. The famous cholesterol "paradoxes" don't overturn that; they mistake a snapshot for the whole film.
Know your apoB (or at least your non-HDL cholesterol), measure your lipoprotein(a) once, and treat a high particle count early rather than late. Diet and exercise help a little; for someone with genuinely high apoB, drugs do most of the work, and the targets and treatment ladder live in lipid management.
Particles, not cholesterol, are what build plaque
Cholesterol itself is not a villain. It is a building block of every cell membrane and the raw material for hormones, vitamin D and bile. The problem is transport: cholesterol does not dissolve in blood, so the body wraps it in particles called lipoproteins, and some of those particles get trapped in artery walls.
A standard lipid panel reports the mass of cholesterol inside those particles: total cholesterol, LDL cholesterol (LDL-C), HDL cholesterol and triglycerides. But plaque is caused by the number of particles that get into the artery wall. Every plaque-building particle (LDL, its precursors VLDL and IDL, and lipoprotein(a)) carries exactly one molecule of apolipoprotein B (apoB), so measuring apoB counts them directly.
ApoB predicts heart risk more reliably than LDL-C: in a pooled analysis of 12 studies covering about 233,000 people, apoB was the strongest of the three markers and LDL-C the weakest.[1] Whether it also beats non-HDL cholesterol, which comes free with any standard panel, is contested. In 94,398 Danish adults the two predicted heart disease about equally well,[2] while in nearly 300,000 UK Biobank adults apoB absorbed the extra risk carried by LDL-C, non-HDL cholesterol and triglycerides.[3]
The cholesterol number and the particle count often disagree. When particles are small, dense and low in cholesterol, the pattern in insulin resistance and metabolic syndrome, LDL-C can look normal while the particle count is high. Across studies, roughly 5% to 25% of people carry more apoB particles than their LDL-C would suggest, most often men and people with metabolic syndrome, diabetes, high triglycerides or low HDL.[4] When the two disagree, risk follows apoB, but the gap is modest: in about 375,000 UK Biobank adults, the roughly 1 in 10 whose apoB ran higher than expected had about 6% to 11% higher risk of heart attack, stroke and similar events.[5] Because apoB tracks the standard numbers closely in most people, whether everyone should be tested is a genuine debate; the case is clearest in metabolic syndrome and diabetes.
One measurement caveat matters now that treatment targets are low. LDL-C on a standard panel is usually calculated, and the old Friedewald formula underestimates it in the low-LDL, high-triglyceride range. Among lab samples with triglycerides of 2.3–4.5 mmol/L (200–399 mg/dL), a Friedewald LDL-C below 1.8 mmol/L (70 mg/dL) was confirmed by direct measurement only about 40% of the time, against about 84% for the newer Martin-Hopkins equation.[6] The 2026 US dyslipidemia guideline prefers Martin-Hopkins or the Sampson (NIH) equation, and says fasting isn't needed for most panels. Fast mainly if your non-fasting triglycerides are 4.5 mmol/L (400 mg/dL) or higher, you have a known or suspected triglyceride disorder, or you have a family history of a lipid disorder or early heart disease.[7] Testing cadence lives in midlife labs.
Plaque builds over decades, so earlier is better
Atherosclerosis is not fat passively clogging a pipe. It is an immune response to apoB particles trapped in the artery wall: the particles oxidise, immune cells move in and gorge on the oxidised fat to become "foam cells," and when those die they form the soft core of a plaque. That ties cholesterol to chronic low-grade inflammation, one of the hallmarks of aging.
Risk is therefore cumulative: plaque burden is roughly the particle count multiplied by the years of exposure. The question is never "is my cholesterol high right now?" but "how high has my particle count been, and for how long?"
This is one of the best-established causal relationships in chronic-disease medicine. A European consensus panel combined more than 200 cohort studies, genetic (Mendelian randomisation) studies and randomised LDL-lowering trials, together covering over two million people, and found the same dose-dependent, straight-line relationship in every design: LDL causes atherosclerosis.[8] The panel's meetings were supported by drug-maker grants, but the genetic arm of the evidence is independent of that funding.
Starting early buys much more. A genetic analysis of more than 300,000 people found that each 1 mmol/L (39 mg/dL) of lifelong lower LDL-C went with about a 55% lower risk of coronary heart disease, roughly three times the benefit per 1 mmol/L when a statin is started in mid or late life.[9] The two figures aren't interchangeable: the genetic one reflects decades of exposure that no pill started late can buy back. Real-world data now point the same way. In Spanish families with inherited high cholesterol (familial hypercholesterolaemia), children treated from a median age of 14.5 had about 40% less lifetime LDL exposure by their 30s than their affected parents, and by age 39, 0.3% had had a cardiovascular event against 5.2% of the parents. This was an observational comparison across generations, so changes in care may explain part of the gap.[10]
Two particle types the LDL number misses
Remnants. The cholesterol inside triglyceride-rich particles as they are stripped down, called remnant cholesterol, is itself a cause of heart disease: in genetic data from more than 70,000 people it raised ischaemic heart disease risk independently of low HDL.[11] Per particle they look roughly four times as damaging as LDL.[12] This is part of why apoB and non-HDL cholesterol beat LDL-C. But what lowers risk is fewer apoB particles, not a lower triglyceride reading. In 10,497 adults with type 2 diabetes, high triglycerides and low HDL, the drug pemafibrate cut triglycerides and remnant cholesterol by about a quarter, apoB rose slightly, and cardiovascular events did not fall.[13] So the target is apoB or non-HDL cholesterol, not a triglyceride number; see dietary fats.
Lipoprotein(a). Lp(a) is an LDL-like particle with an extra protein tail; it causes plaque and inflammation and, particle for particle, is more harmful than ordinary LDL. Roughly one in five people inherit high levels. It is more than 90% set by genes and changes little after childhood, though it tends to rise in women at menopause; diet shifts it by only 10–15%, and exercise and weight loss barely move it.[14] Statins don't lower it and may nudge it up by about 1 mg/dL, which is no reason to stop one. Below about 75 nmol/L (30 mg/dL) is low-risk; at 125 nmol/L (50 mg/dL), a level about one in five people reach, risk is roughly 40% higher; at about 430 nmol/L (180 mg/dL), about four times as high. Risk rises smoothly, so these are convenient markers rather than true thresholds. US guidelines now recommend that every adult have it measured at least once, and that a high result mean lowering LDL-C harder and managing every other risk factor more tightly.[15]
Drugs designed to lower Lp(a) cut it by roughly 70% to nearly 100% in earlier trials, but whether that prevents heart attacks is unanswered. The first to report, pelacarsen, lowered Lp(a) but did not significantly reduce cardiovascular events in 8,323 people with established cardiovascular disease and high Lp(a), according to topline results in September 2026.[16] None is approved.
Lowering apoB prevents heart attacks
This is the payoff, and the trial evidence is strong. Pooling statin trials that lasted about five years, each 1 mmol/L (39 mg/dL) drop in LDL-C cut the yearly rate of heart attacks, strokes and artery-opening procedures by just over a fifth, with no lower threshold at which the benefit ran out and no excess of cancer or non-vascular death.[17]
Non-statin drugs that clear apoB particles confirm the principle. Adding ezetimibe, which blocks cholesterol absorption, cut events further.[18] The injectable PCSK9 inhibitors reduced major events by about 15% in people with established heart or artery disease,[19] and in 12,257 people with narrowed arteries or high-risk diabetes who had never had a heart attack or stroke, evolocumab cut coronary deaths, heart attacks and ischaemic strokes by about a quarter over a median of 4.6 years.[20] Bempedoic acid, a pill for people at high heart risk who can't take statins because of side effects, cut major heart events by about 13% in nearly 14,000 such people, but did not measurably reduce deaths and slightly raised gout and gallstones.[21] Inclisiran, a twice-yearly injection, reliably halves LDL-C, but its outcome trials are still running.[22]
Not every LDL-lowering drug pays off: the CETP inhibitors (below) are the exception that shows the principle is about clearing apoB particles. The pivotal drug trials are all industry-funded; the strongest funding-independent support is the genetic and academic meta-analytic work above. Drug choice and dosing live in lipid management.
Statins: what the side effects really are, and the age question
Muscle aches are the most-cited reason people quit statins, but in large blinded trials they were almost as common on placebo. Across 19 trials with about 124,000 people, 27.1% on a statin and 26.6% on placebo reported muscle pain or weakness; more than 90% of complaints among statin takers were not caused by the drug, and the small real excess was mostly mild and in the first year.[23] In one small trial in people who had already given up statins, a dummy pill produced about 90% of the extra symptoms the statin did.[24] The symptoms are real; the molecule is usually not the cause.
New-onset diabetes is a genuine drug effect. Statins raised the chance of a diabetes diagnosis by about 10% on low- or moderate-intensity doses and more than a third on high-intensity doses, reflecting a small rise in blood sugar, with most new diagnoses in people already close to the threshold. That harm is already counted in the trials' overall drop in heart attacks and strokes, so the benefit outweighs it.[25]
Very low LDL and the brain is a worry that hasn't shown up. In an uncontrolled extension of a trial, 473 people who took evolocumab for a median of about five more years at a median LDL-C of about 0.9 mmol/L (35 mg/dL) showed no decline in thinking scores, though this still needs testing in people at higher risk of dementia,[26] and pooled blinded statin trials found no excess of cognitive impairment.[27] Nor has excess cancer, though trial follow-up is measured in years, not decades.
Age. Statins clearly help for primary prevention up to about 75. Above 70 the trial evidence is mixed: in 9,971 healthy adults aged 70 or older, atorvastatin cut heart attacks, strokes and related events by about 30% over a median of 5.9 years, but did not help people live longer free of dementia or disability, and muscle, liver and diabetes-related problems were more frequent on the statin.[28] The useful lens is time-to-benefit: in adults aged 50–75 it takes about 2.5 years of treatment to prevent one cardiovascular event per 100 people, and only one of eight trials found fewer deaths from any cause.[29] A primary-prevention statin makes sense when remaining life expectancy comfortably exceeds that, which is also why stopping one becomes reasonable in frailty or limited life expectancy.
The three paradoxes, and why none overturns this
Each is real and worth understanding. In each, a snapshot or a marker is being mistaken for lifelong cumulative exposure.
Very high HDL is not protective. In large observational studies, death from any cause follows a U-shaped curve: both low and very high HDL go with higher mortality, though the high end is less clear-cut than often claimed. In two Copenhagen cohorts of about 116,000 adults, men at 3.0 mmol/L (116 mg/dL) or above had about twice the death rate of men at the lowest-risk level, while in women a clear excess appeared only at 3.5 mmol/L (135 mg/dL) or above.[30] Very high HDL often signals dysfunctional particles, or heavy drinking, which carries its own mortality. The decisive evidence is from drugs: the CETP inhibitors raised HDL dramatically and mostly failed to help even though some also cut LDL; torcetrapib raised the risk of death by about 58%, possibly through an off-target rise in blood pressure the investigators suspected.[31] The one drug in the class that modestly helped, anacetrapib, did so in line with its fall in non-HDL cholesterol. HDL is a marker of metabolic state, not a lever, and the old idea that alcohol protects the heart by raising HDL has collapsed under genetic analysis.
Low cholesterol in the very old. In people in their late 80s and 90s, higher cholesterol often goes with lower death rates over the next few years; among 168 Sardinian nonagenarians, those with higher LDL-C lived about a year longer.[32] The evidence is small studies built on a single late-life reading. The usual explanation, most likely but not settled, is reverse causation: illness, frailty, malnutrition and hidden cancer all lower cholesterol. Cholesterol still matters at that age: in healthy Danes aged 80–100, LDL-C of 5 mmol/L (193 mg/dL) or more meant about three times the heart-attack risk of LDL-C below 3 mmol/L (116 mg/dL).[33] And genetic studies, which capture a lifetime rather than one frail year, link higher LDL-C to a shorter life: about 1.2 years per standard deviation.[34] A widely shared review arguing the pattern disproves the cholesterol story [35] had serious methodological flaws in an independent Oxford appraisal and cannot outweigh that evidence.[36]
Keto and the lean-mass hyper-responder. Some lean, fit people on a ketogenic diet develop very high LDL-C (often above 4.9 mmol/L, or 190 mg/dL), high HDL and low triglycerides. The reassuring evidence is thin and preliminary. In one snapshot scan, 80 hyper-responders with LDL-C around 7.0 mmol/L (272 mg/dL) had no more coronary plaque than matched people at about 3.2 mmol/L (123 mg/dL), but both groups had very little plaque and one scan can't show growth.[37] A follow-up claiming plaque growth didn't depend on apoB was retracted in 2026 for errors in its methods.[38] Across randomised trials, ketogenic diets raised LDL-C by about 0.4 mmol/L (17 mg/dL) on average against low-fat diets, even in people who are not hyper-responders.[39] There are no outcome data for hyper-responders. The honest position is uncertainty: hyper-responders who want to stay in ketosis without gambling on long-term risk can lower apoB with ezetimibe.
What diet, exercise and sleep can do
Lifestyle reliably improves the lipid profile, but the effects are modest. The levers still matter because they act for decades and improve much more than lipids.
- Diet. The main lever is substitution. In randomised trials in 53,758 adults, cutting saturated fat for at least two years modestly lowered LDL-C and led to about 17% fewer cardiovascular events, though not clearly fewer deaths, and the trials suggested a benefit whether the saturated fat was replaced with polyunsaturated fat or with carbohydrate. The trials did not separate refined from whole-food carbohydrate.[40] Dietary cholesterol matters less than once thought: eating more than today's average raises LDL-C modestly in feeding studies, but large observational studies generally find no link with heart disease.[41] Detail lives in dietary fats, dietary patterns and eggs.
- Exercise. Across 148 randomised trials, exercise lowered LDL-C by only about 0.19 mmol/L (7 mg/dL) and raised HDL slightly; aerobic and combined training lowered LDL-C, while resistance training on its own mainly raised HDL.[42] In a one-year trial in 188 overweight men, endurance training lowered apoB even though LDL-C did not change significantly.[43] Long spells of sitting suppress the muscle enzyme that clears triglycerides.
- Sleep and shift work. The links are weak and observational. In a large Chinese cohort, sleeping 9 hours or more was linked to later low HDL, but not to high LDL-C,[44] and night-shift miners had slightly higher total cholesterol than day workers in a one-time snapshot.[45] A regular circadian rhythm is a sensible habit rather than a proven lipid treatment.
- Stress. Acute stress temporarily raises cholesterol and triglycerides through stress hormones, and structured stress reduction modestly improves the profile.
Where a calcium scan fits
A coronary artery calcium (CAC) scan counts established, calcified plaque. It is a strong predictor and can sharpen the decision about who most needs treatment: in one large observational study, statin users with a score of zero did no better than non-users over about 10 years, while above 100, treating about 12 people for a decade prevented one heart attack, stroke or cardiovascular death, though healthier habits among statin takers may explain part of that.[46] Randomised trials of scan-guided care have not yet shown fewer heart attacks or deaths.[47]
A zero is not a clean bill of health. It misses the soft, early plaque this page describes, and it should not be used to withhold a statin from people over 40 with diabetes, current smokers, people with familial hypercholesterolaemia or LDL-C above 4.9 mmol/L (190 mg/dL), or those with a strong family history of early heart disease. It also wears off: about half of people with an initial zero had some calcium within 10 years, and about one in five of the coronary events they had over that decade struck before a repeat scan at 3–5 years.[48] Treat it as a tie-breaker for apoB-based decisions, not a substitute.
What's overrated
- Fixating on a single LDL-C reading. One number on one day says little; lifelong apoB exposure is what builds plaque.
- Dietary-cholesterol panic. Eggs and shellfish raise blood cholesterol only modestly; the overall dietary pattern and saturated-fat substitution matter more.
- Reading the paradoxes as a green light. They are nuances at the edges, not licence for a healthy midlife adult to ignore a high apoB.
Further reading
- Sniderman AD, et al. A meta-analysis of low-density lipoprotein cholesterol, non-high-density lipoprotein cholesterol, and apolipoprotein B as markers of cardiovascular risk. Circ Cardiovasc Qual Outcomes 2011.[49]
- Ference BA, et al. 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. Eur Heart J 2017.[50]
- Ference BA, et al. 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. J Am Coll Cardiol 2012.[51]
- Cholesterol Treatment Trialists' (CTT) Collaboration; Baigent C, et al. Efficacy and safety of more intensive lowering of LDL cholesterol: a meta-analysis of data from 170,000 participants in 26 randomised trials. Lancet 2010.[52]
- Varbo A, et al. Remnant cholesterol as a causal risk factor for ischemic heart disease. J Am Coll Cardiol 2013.[53]
- Cholesterol Treatment Trialists' Collaboration. Effect of statin therapy on muscle symptoms: an individual participant data meta-analysis of large-scale, randomised, double-blind trials. Lancet 2022.[54]
- Wood FA, et al. N-of-1 trial of a statin, placebo, or no treatment to assess side effects. N Engl J Med 2020 — the SAMSON trial.[55]
- Yourman LC, et al. Evaluation of time to benefit of statins for the primary prevention of cardiovascular events in adults aged 50 to 75 years: a meta-analysis. JAMA Intern Med 2021.[56]
- Zoungas S, et al. Atorvastatin, cardiovascular events, and disability-free survival in older adults. N Engl J Med 2026.[57]
- Bohula EA, et al. Evolocumab in patients without a previous myocardial infarction or stroke. N Engl J Med 2026.[58]
- Hooper L, et al. Reduction in saturated fat intake for cardiovascular disease. Cochrane Database Syst Rev 2020.[59]
- Budoff M, et al. Carbohydrate restriction-induced elevations in LDL-cholesterol and atherosclerosis: the KETO trial. JACC Adv 2024.[60]