Cholesterol and lipoproteins

Heart disease is the single biggest barrier to a long life, and the cholesterol that drives it is widely misunderstood: what ages your arteries is not the number on one panel but the lifelong number of cholesterol-carrying particles your blood vessels have been exposed to. The famous cholesterol "paradoxes" don't overturn that — they just confuse people who read a snapshot as if it were the whole movie.

Cholesterol itself is not a villain — it is an essential 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. Some of those particles get trapped in artery walls and start the slow, decades-long process of atherosclerosis. Understanding which particles matter, and why, explains nearly every cholesterol-related recommendation on the rest of this site — and dissolves most of the apparent contradictions that circulate online.

Particles, not cholesterol — why apoB is the real signal

Strong. The standard lipid panel reports the mass of cholesterol carried inside particles: total cholesterol, LDL cholesterol (LDL-C), HDL cholesterol, and triglycerides. But atherosclerosis is not caused by cholesterol mass — it is caused by the number of artery-penetrating particles. Every atherogenic particle (LDL, its precursors VLDL and IDL, and lipoprotein(a)) carries exactly one molecule of apolipoprotein B (apoB) on its surface. Measuring apoB therefore counts those particles directly, and large analyses rank it above non-HDL cholesterol, with LDL-C the least reliable of the three.^[1]

This matters because the cholesterol number and the particle count often disagree. When particles are small, dense, and cholesterol-depleted — the pattern seen in insulin resistance and metabolic syndrome — LDL-C can look reassuringly normal while the particle count is dangerously high. In large primary-prevention cohorts, when LDL-C and apoB disagree, cardiovascular risk tracks apoB, not LDL-C.^[2] In nearly 300,000 UK Biobank adults, once apoB was in the model the residual risk carried by LDL-C, non-HDL-C, and triglycerides all vanished — apoB captured what they were each pointing at.^[3] Discordance is much more common when triglycerides are high (around 40% of people versus roughly 22% otherwise), which is why apoB earns its place most clearly in metabolic syndrome and diabetes. That said, because apoB correlates almost perfectly with the standard numbers in most people, whether universal apoB testing is worth the added cost and clinician education is a genuine ongoing debate rather than a settled point. Strong that apoB is the single most accurate atherogenic-particle marker; moderate/contested on testing everyone.

One measurement caveat matters now that treatment targets are so low. LDL-C on a standard panel is usually calculated, not measured, and the old Friedewald formula systematically underestimates it precisely in the low-LDL, high-triglyceride range where decisions are now made. The newer Martin-Hopkins method is more accurate there and is guideline-endorsed.^[4] Fasting is no longer required for a routine panel, and a non-fasting sample actually captures the remnant particles discussed below a little better. The practical targets, testing cadence, and treatment ladder live in lipid management and midlife labs; this page is about the underlying biology.

The biology: retention and cumulative exposure

Strong. Atherosclerosis is not fat passively clogging a pipe. It is an immune-inflammatory response to apoB particles becoming trapped in the artery wall. Once retained beneath the vessel lining, the particles oxidise, recruit immune cells (monocytes that mature into macrophages), and those macrophages engorge themselves on the oxidised lipid to become "foam cells." Foam cells die, forming the soft necrotic core of a plaque. This connects cholesterol directly to one of the integrative hallmarks of aging — chronic, low-grade inflammation.

The single most important consequence is that risk is cumulative. Each particle that crosses the artery wall has some probability of being trapped, so total plaque burden is the particle count multiplied by the years of exposure — an area under the curve, not a snapshot. This is why the question is never "is my cholesterol high right now?" but "how high has my particle count been, for how long?"

This is not just a marker correlation — it is one of the best-evidenced causal relationships in all of chronic-disease medicine. A European consensus panel triangulated meta-analyses of more than 200 prospective cohort studies, genetic (Mendelian randomization) studies, and randomised LDL-lowering trials — together covering over two million people — and found the same dose-dependent, straight-line relationship in every design, concluding that LDL does not merely track atherosclerosis but causes it. Strong / established consensus.^[5] (The panel's meetings were supported by unrestricted grants from drug makers and several authors report industry ties, but the genetic arm of the evidence is funding-independent.)

It is also why lowering apoB earlier in life buys far more than the same reduction started in late middle age. A genetic analysis of more than 300,000 people found that each 1 mmol/L (39 mg/dL) lifelong lower LDL-C was tied to a 55% lower risk of coronary heart disease — roughly triple the effect of the same drop achieved by starting a statin in mid or late life, which the trial evidence puts at about a 22% lower risk per 1 mmol/L per year of treatment.^[6] The two numbers are not interchangeable: the genetic figure reflects decades of exposure and cannot be achieved by a pill started late — but it is the strongest available argument that "earlier is better." Strong (genetic), with that interpretive caveat.

Remnant cholesterol: the residual risk in triglyceride-rich particles

Strong (genetic + epidemiology). Not all atherogenic particles are LDL. The cholesterol carried inside triglyceride-rich VLDL and IDL "remnants" — what's left as those particles are stripped down — is itself a causal driver of heart disease. A large genetic study in more than 70,000 people found that a rise in remnant cholesterol raised ischaemic heart disease risk independently of low HDL, and unlike LDL these remnants may not even need to be oxidised to be swallowed by artery-wall macrophages.^[7] This is part of why apoB and non-HDL cholesterol outperform LDL-C: they count remnant particles that an LDL-C measurement misses. The important nuance is that the causal signal is for the remnant particles, not the triglyceride number as such — trials that lower triglycerides directly have given mixed results, so the target is apoB/non-HDL-C, not chasing a triglyceride reading (covered in dietary fats).

Lipoprotein(a): the genetic wildcard

Strong / causal. Lipoprotein(a), written Lp(a), is an LDL-like particle with an extra protein tail (apolipoprotein(a)) that structurally resembles plasminogen, the molecule that dissolves clots. That makes Lp(a) simultaneously atherogenic, pro-thrombotic, and pro-inflammatory — meaningfully more dangerous, particle for particle, than ordinary LDL. Roughly one in five people inherit high levels, which are set genetically, stable for life from early childhood, and essentially unmoved by diet, exercise, weight loss, or statins.

As a rough guide, a level below about 30 mg/dL (75 nmol/L) is low-risk, above roughly 50 mg/dL (125 nmol/L) is high-risk, and very high levels — above about 180 mg/dL (450 nmol/L) — carry risk on a par with the inherited cholesterol disorder familial hypercholesterolaemia. But the underlying relationship is continuous, so these cut-offs are a clinical convenience rather than true thresholds.

Because it is invisible to lifestyle change and to a standard panel, the recommendation is simple: measure it once in a lifetime to know whether you carry this risk. If it is high, the response for now is to intensify every other modifiable factor. Drugs that specifically lower Lp(a) — antisense and small-interfering-RNA injectables that cut it by 80–95%, and an oral agent in earlier testing — are in late-stage outcome trials, but the pivotal question of whether lowering Lp(a) actually prevents heart attacks is still unanswered, with the first hard-outcome results expected from 2026. Causality: strong; the therapies: promising for lowering Lp(a), unproven for outcomes. None is approved yet. What to do about an elevated result is covered in lipid management.

The paradoxes that confuse people

Three findings get cited as if they overturn the apoB story. None of them do — but each is real and worth understanding, because the explanation is the same in all three: a single late-life or short-term snapshot is not the same thing as lifelong cumulative exposure.

Very high HDL is not protective

Moderate (observational). HDL was long called "good cholesterol" because it carries cholesterol away from tissues back to the liver. The reality is a U-shaped curve: both low and very high HDL associate with higher mortality. Cardiovascular and all-cause death rise once HDL cholesterol exceeds roughly 2.3 mmol/L (90 mg/dL) in men and about 2.8–3.0 mmol/L (110–116 mg/dL) in women.^[8] The extra risk at the high end appears in non-diabetic men too, independent of the usual confounders.^[9]

Very high HDL often signals dysfunctional particles that have lost their anti-inflammatory and cholesterol-offloading capacity, or is driven by heavy alcohol intake (which carries its own mortality). Crucially, drugs that raise HDL have repeatedly failed to reduce events in randomised trials — a whole class of HDL-raising agents (the CETP inhibitors) provides the falsification. Torcetrapib raised HDL by 72% yet was stopped early because it increased deaths, through an off-target effect on blood pressure.^[10] Evacetrapib more than doubled HDL and also cut LDL, yet was halted for futility with no effect on events at all.^[11] Where a later drug in this class did modestly reduce events, the benefit tracked its LDL/apoB lowering, not the HDL rise. The lesson: HDL is a marker of metabolic state, not a lever to pull — and the old idea that alcohol's HDL-raising effect protects the heart has collapsed under genetic analysis.

Low cholesterol in the very old

Moderate — reverse causation. In people in their late 80s, 90s, and beyond, higher total cholesterol and LDL-C often associate with lower mortality — the opposite of the relationship in midlife. This "cholesterol paradox" shows up even in longevity hot spots such as Sardinia.^[12] The dominant explanation is reverse causation: serious illness, frailty, malnutrition, and occult cancer all lower cholesterol, so low cholesterol in the very old is frequently a marker of approaching decline rather than its cause.^[13]

A widely circulated review has been used to argue this pattern disproves the cholesterol story altogether — reporting that in most cohorts of people over 60, higher LDL-C went with lower mortality.^[14] But independent reviewers flagged serious weaknesses: no pre-registered protocol, a literature search that missed most relevant cohorts, reliance on pooled rather than individual data, and — decisively — no accounting for the reverse causation that dominates in the very old. It cannot outweigh the genetic and trial evidence.

This does not contradict the lifelong story. Genetic studies — which capture decades of exposure rather than one frail year — consistently show that a lifelong predisposition to lower LDL-C and apoB extends lifespan and lowers cardiovascular risk.^[15] Low cholesterol in a 95-year-old can mean illness; low cumulative apoB across a lifetime means protected arteries. They are not the same measurement. See Blue Zones for how to read longevity-population data without over-fitting.

Keto and the Lean-Mass Hyper-Responder

Weak / preliminary — flag the uncertainty. A specific group of lean, fit, metabolically healthy people who adopt a ketogenic (very-low-carbohydrate) diet develop a striking triad: very high LDL-C (often above 4.9 mmol/L / 190 mg/dL, sometimes far higher), high HDL, and low triglycerides. These are the "Lean-Mass Hyper-Responders." The proposed mechanism — the Lipid Energy Model — is that with carbohydrate scarce, a lean liver ships large amounts of fat to muscle for fuel via triglyceride-rich particles that remodel into many LDL particles; the leaner the person, the higher the LDL spike.

The provocative data come from the KETO-CTA study: hyper-responders maintaining a mean LDL-C around 7.0 mmol/L (272 mg/dL) for several years showed no greater coronary plaque burden than a comparison group with far lower LDL-C, and over one year of follow-up the change in plaque was predicted by existing plaque, not by apoB or LDL-C — summarised as "plaque begets plaque, but apoB does not" in this phenotype.^[16] That is genuinely surprising and an active area of debate.^[17] But the cohorts are young, the windows are short (a few years against a disease that takes decades), and no long-term outcome data exist yet. The honest position is uncertainty, not reassurance: hyper-responders who want to stay in ketosis but not gamble on unproven long-term risk can lower apoB with a cholesterol-absorption inhibitor (ezetimibe).

What actually moves your lipids

Moderate. Lifestyle reliably improves the lipid profile, but the effect sizes are modest — for someone with genuinely high apoB, pharmacology does the heavy lifting. The lifestyle levers still matter because they act for decades and improve much more than lipids.

Diet. The dominant dietary lever is substitution: replacing saturated fat with polyunsaturated fat (and high-quality carbohydrate) lowers LDL-C and cuts cardiovascular events by roughly 17% in pooled trials, whereas swapping it for refined carbohydrate does little.^[18] Dietary cholesterol itself (eggs, shellfish) has a weak effect in most people, because the liver compensates — modern guidelines dropped the old numeric limit.^[19] Detail lives in dietary fats and dietary patterns.
Exercise. Aerobic training shifts the whole panel modestly and, more importantly, lowers apoB and improves particle quality even when LDL-C barely moves.^[20] Resistance training mainly lowers triglycerides; combining the two is best.^[21] Even prolonged sitting matters — it collapses the muscle enzyme (lipoprotein lipase) that clears triglycerides from the blood.
Sleep and circadian rhythm. Lipid abnormalities follow a U-shaped curve with sleep duration, worst at both short and long ends.^[22] Night-shift work independently raises total cholesterol, even in people who report sleeping well — keeping a regular circadian rhythm is part of lipid metabolism, not separate from it.^[23]
Stress. Acute psychological stress transiently raises cholesterol and triglycerides through the stress-hormone axis, and chronic stress sustains the pattern; structured stress reduction modestly improves the profile.

How clinicians turn all of this into a personal plan — formal cardiovascular-risk scoring (such as Europe's SCORE2, which estimates 10-year risk of fatal and non-fatal events), apoB targets by risk tier, and when drugs are warranted — sits in lipid management. The unifying principle is "lower and earlier is better" for apoB across the whole adult lifespan.

Does lowering apoB actually help? The outcome evidence

Strong. The whole page rests on the claim that apoB builds plaque; the payoff is that lowering it prevents events — and by more or less any mechanism. Statins are the anchor: pooling data from many trials, each 1 mmol/L (39 mg/dL) drop in LDL-C lowered major vascular events by just over a fifth per year of treatment, with no lower threshold at which the benefit ran out and no excess of cancer or non-vascular death.^[24] That "lower is better, by any mechanism" principle is confirmed by non-statin drugs: adding ezetimibe, which blocks cholesterol absorption, cut events further,^[25] and the injectable PCSK9 inhibitors, which push LDL-C to very low levels, reduced major events by about 15%.^[26] A second PCSK9 trial showed a nominal reduction in death, though the trials were short (two to three years), which likely limits how large a mortality signal they can show.^[27] Bempedoic acid, an oral option for people who cannot tolerate statins, reduced events too — including, notably, in a primary-prevention group with no prior heart attack.^[28] Newer agents such as the twice-yearly injectable inclisiran reliably lower LDL-C by about half, but their outcome trials are still running, so they remain emerging for hard endpoints.^[29] All of these pivotal trials are industry-funded; the strongest funding-independent support comes from the genetic and academic meta-analytic work above. Detailed dosing and choice of agent live in lipid management.

Two safety questions dominate the public debate and deserve a straight answer. Muscle aches are the most-cited reason people quit statins, but blinded "n-of-1" trials show most of that burden is not caused by the drug: in one crossover study, taking a dummy pill produced about 90% of the symptom load that the statin did, and a larger series found no difference at all between statin and placebo months.^[30]^[31] This is the nocebo effect — the symptoms are real, but the molecule is usually not the cause. New-onset diabetes is a genuine drug effect: statins raise the risk modestly (roughly 10% on lower doses, about a third on high doses), concentrated in people already close to the diabetes threshold and mostly reflecting small blips in blood sugar rather than new disease; the vascular benefit outweighs it.^[32] A parallel worry — that very low LDL-C harms the brain or raises cancer risk — has not shown up: dedicated cognitive substudies found no harm even at LDL-C below 25 mg/dL, and the trial meta-analyses found no excess cancer, though follow-up is measured in years, not decades. Strong (no harm signal), tempered by finite follow-up.

For a longevity reader, the sharpest question is when. Statins clearly help for primary prevention up to about age 75; beyond that the randomised evidence thins out and two large trials in older adults are still running. The useful lens is time-to-benefit: a meta-analysis of more than 65,000 adults aged 50–75 estimated it takes about 2.5 years of treatment to prevent one cardiovascular event per 100 people, with no mortality benefit visible inside that window.^[33] The practical rule follows directly: a primary-prevention statin makes sense when remaining life expectancy comfortably exceeds the time-to-benefit — which is also why deprescribing becomes reasonable in frailty or limited life expectancy. Strong through 75; uncertain beyond.

Where the CT scan fits: coronary artery calcium

Strong observational (no randomised trial of scan-guided treatment). A coronary artery calcium (CAC) scan counts established, calcified plaque, and it can sharpen the decision of who most needs lipid-lowering. In a large observational study, statins reduced events in people with any calcium but not in those with a score of zero — where the number needed to treat over a decade was in the thousands versus roughly a dozen at high scores.^[34] But a zero score is not a clean bill of health: it measures only the hardened, late-stage plaque, not the soft, lipid-rich early lesions this page describes. It is less reassuring in younger people and in high-Lp(a) carriers, a meaningful share of zero scores turn positive within a few years, and a portion of events still occur in people who scored zero. Treat it as a tie-breaker that refines apoB-based decisions, not a substitute for them.

What's overrated

Chasing a high HDL. It is a marker, not a target; very high levels can signal dysfunction, and HDL-raising drugs don't improve outcomes.
Fixating on a single LDL-C reading. One number on one day says little. Lifelong apoB exposure is the quantity that builds plaque.
Dietary-cholesterol panic. For most people, eggs and shellfish move blood cholesterol little; the dietary pattern and saturated-fat substitution dominate.
Reading the paradoxes as a green light. Low cholesterol in the frail elderly and the keto hyper-responder findings are interesting nuances at the edges — not licence for a healthy midlife adult to ignore a high apoB.