Cholesterol.

The artery is lined by a thin active layer, the endothelium. Where blood flow is turbulent (branch points, curves) it becomes leaky and inflamed, and ApoB-carrying particles (LDL and its relatives) pass into the wall and get trapped — the 'response-to-retention' model, the initiating step. Trapped particles are oxidised, which the body reads as danger.

Oxidised particles recruit immune cells (monocytes) that become macrophages, which gorge on the trapped cholesterol and turn into foam cells — the earliest visible lesion (a 'fatty streak', present in many people by their 20s). Foam cells die, spilling a soft lipid/necrotic core; smooth-muscle cells lay a fibrous cap over it. The whole thing is self-amplifying: more particles in the blood means more retained, more inflammation, more plaque. Evidence ●●●●●

Soft (non-calcified) plaque is lipid-rich, inflamed, thin-capped — the young, dangerous, rupture-prone form (it shows as 'low-attenuation' plaque on a CT scan). Dense calcified plaque is older and largely a healing/stabilising signature — mineralised, quieter, less likely to rupture. A key nuance: spotty micro-calcification flags active instability, while large confluent dense calcium marks stability. This distinction runs through the whole imaging story.

The lesion that kills is usually not the tightest narrowing. It's the thin-cap fibroatheroma — a large soft lipid core under a fibrous cap thinner than 65 microns, riddled with inflammation. Under stress the cap tears, the core hits flowing blood, and a clot forms in minutes. Because these plaques are soft rather than bulky, they are often non-obstructive — they barely narrow the artery, so they look reassuring on an angiogram or stress test. Evidence ●●●●●

Roughly 60–70% of heart attacks arise from lesions narrowing the artery less than 50%. The landmark PROSPECT study found future events were predicted by total plaque burden, a small lumen, and a thin-cap fibroatheroma — not by the tightest stenosis. This is the single most important reframing in the field: how much plaque you have, and how soft it is, matters more than the 'percent narrowing' — and it's exactly why a calcium score of zero can miss risk (covered in the scans section).

Every atherogenic particle — LDL, VLDL, IDL, remnants, and Lp(a) — carries exactly one ApoB molecule. So an ApoB level is a direct count of the particles that lodge in the artery wall. LDL-cholesterol measures the cholesterol cargo inside those particles, which varies; ApoB counts the trucks. Atherosclerosis tracks the trucks — which is why ApoB is the more accurate number (test section).

Causality rests on three lines that all give the same log-linear, dose-dependent answer. Genetics (Mendelian randomisation): 50+ different LDL-lowering gene variants each lower heart disease by the same amount per unit of LDL — whatever the mechanism — the fingerprint of a causal agent. Trials (CTT meta-analyses): each 1 mmol/L (~39 mg/dL) lower LDL cuts major vascular events ~22%, with benefit continuing safely to very low levels and no threshold. Epidemiology: a strong graded association across ~2 million people. The European consensus verdict: LDL causes atherosclerotic cardiovascular disease. Evidence ●●●●●

Arterial damage is a function of particle concentration times duration — think 'cholesterol-years', like pack-years for smoking. That's why lifelong genetically lower LDL cuts heart disease roughly three times more per unit than the same reduction started in middle age. The practical corollary is lower and earlier: the area under the curve over decades is what builds plaque.

This exact claim — that ApoB/LDL causes disease rather than just marking it — is the precise point a minority of statin skeptics dispute. It's a real debate about the evidence and it's handled fairly in the statins section; the mainstream position, backed by the Mendelian-randomisation data above, is that causality is settled.

Serial ultrasound-inside-the-artery (IVUS) trials line up as a dose-response ladder measuring percent atheroma volume (PAV). REVERSAL: LDL 110 progressed (+2.7%), LDL 79 halted. ASTEROID: LDL 61 regressed plaque (−0.98%) — the first clear reversal. SATURN: both maximal statins regressed (~−1%). GLAGOV: adding a PCSK9 inhibitor drove LDL to 37 mg/dL and regressed plaque in 64% of patients, with benefit continuing down to LDL ~20 and no floor. Evidence ●●●●○

Shrinking volume is only half of it. In the HUYGENS and PACMAN-AMI trials, intensive lowering with a PCSK9 inhibitor roughly doubled fibrous-cap thickening, shrank the lipid core, and cut plaque inflammation — remodelling plaque toward a stable, rupture-resistant form. Evidence ●●●●○

Counterintuitively, statins increase dense coronary calcium even as they regress plaque and cut events. This isn't harm — it's the soft, inflamed plaque converting to dense, quiet calcium. So a rising calcium score on a statin is usually stabilisation, not failure — which is why calcium density, not just the total score, now matters.

Net regression tends to begin once LDL is pushed below ~70 mg/dL, and it keeps improving with no clear lower limit. But plaque changes are slow — every trial ran 18–24+ months to shift PAV ~1% — so this is a matter of years of sustained low ApoB, not a quick fix, and stopping reverses the gains. The imaging changes are surrogates, but they're corroborated by hard-outcome trials (FOURIER) using the same drugs.

LDL-C: 'optimal' <100 mg/dL (<2.6 mmol/L) for most, but the risk-based targets go lower — <70 (<1.8) for high risk, <55 (<1.4) for very-high risk. Triglycerides: normal <150 (<1.7); optimal <100 (<1.1), ideal <70. HDL: the 40–60 range is the reassuring zone — both very low and very high (>80) associate with higher mortality (a U-curve), and it's a marker, not a lever. Non-HDL (total minus HDL) captures the cholesterol in all atherogenic particles; target = your LDL goal + 30.

Two useful numbers hide in every panel at no extra cost. Total/HDL <3.5 is a strong overall marker. Triglyceride/HDL (in mg/dL) is a cheap insulin-resistance and small-dense-LDL proxy — <2 optimal, >3 flags likely insulin resistance (caveat: less valid in Black patients, who run lower triglycerides). And remnant cholesterol = total − LDL − HDL: ideally <24 mg/dL — a causal, atherogenic fraction that standard LDL misses.

LDL is usually calculated, not measured (the Friedewald equation), and it underestimates LDL when triglycerides are up — labelling people 'at goal' when they aren't. When triglycerides are elevated, prefer the Martin-Hopkins calculation, a directly-measured LDL, or simply read non-HDL cholesterol (which needs no assumption). Evidence ●●●●○

ApoB directly counts the atherogenic particles (one ApoB each), so it beats LDL-C and non-HDL-C as a risk marker — superior in 9 of 9 head-to-head discordance studies, and the metric to which clinical benefit is proportional. It shines exactly where LDL-C misleads: insulin resistance, high triglycerides, small dense LDL, diabetes — where LDL can read 'normal' while particle number (and true risk) is high. Targets: <90 desirable, <80 optimal, <65 for very-high risk (mg/dL). A serious expert case says it should replace LDL-C outright; guidelines currently call it a superior add-on. Evidence ●●●●●

Lp(a) is an LDL-like particle with an extra prothrombotic tail. It's ~90% genetically fixed, essentially lifelong, and elevated in ~20% of people (~1.4 billion) — an independent, causal risk factor for both heart disease and aortic valve stenosis. Because it's stable, you measure it once: desirable <30 mg/dL (<75 nmol/L), high risk >50 mg/dL (>125 nmol/L). Prefer the nmol/L assay (the two units don't convert cleanly). The catch: statins, diet, and exercise barely move it — so an elevated Lp(a) is a reason to drive everything else (ApoB, blood pressure) harder, and targeted drugs are only now in trials. Evidence ●●●●○

hs-CRP (vascular inflammation: <1 low / 1–3 average / >3 high) is a genuine, guideline-endorsed risk-enhancer at ≥2 mg/L — and CANTOS proved inflammation is a causal, targetable axis independent of LDL. Metabolic markers — fasting glucose, HbA1c, fasting insulin, and the TG/HDL ratio — capture insulin resistance, the upstream driver of atherogenic dyslipidaemia and the one you can most modify. These, plus free remnant cholesterol, are the additions with real yield. Evidence ●●●●○

Several markers are prognostic but not actionable, and lowering them didn't help in trials: LDL particle number (NMR) is largely redundant with ApoB (and arguably inferior on the newest data); oxidised LDL and myeloperoxidase have no standardised assay or treatment; Lp-PLA2 failed as a drug target (darapladib trials were null); and homocysteine is the textbook case — easy to lower with B-vitamins, but doing so didn't reduce events (HOPE-2, NORVIT). Interesting biology; not decision-changing. Evidence ●●○○○ as routine tests

Most people capture ~90% of the value from a short set: a standard panel (which hands you non-HDL and remnant cholesterol free), one ApoB, a one-time Lp(a), glucose/HbA1c, and — if your calculated risk is intermediate and the decision is uncertain — a coronary calcium score. The big direct-to-consumer 'advanced cardiac panels' bundle many markers, look impressive, and rarely change what a clinician would do. More tests mostly buy anxiety and cost.

Each turns age, sex, blood pressure, lipids, smoking, and diabetes into a 10-year risk percentage. The US Pooled Cohort Equations were standard; the AHA's new PREVENT equations (2023) added kidney and metabolic factors, dropped race, and extended to 30-year risk — and in doing so roughly halved estimated risk for many people, which by one analysis would make ~40 million fewer Americans statin-eligible. The UK uses QRISK3. Rough bands: <5% low, 7.5–20% intermediate (the decision zone), >20% high.

Age dominates the math. Most men over 65 clear the threshold almost regardless of their lipids, while a 35-year-old with a terrible ApoB and sky-high Lp(a) can score 'low' because they're young — even though their lifetime risk is severe. The scores also ignore Lp(a), the depth of your family history, and whether you already have plaque. A '10% 10-year risk' is a population frequency, not your personal fate. Excellent for triage; poor as a diagnosis.

Guidelines patch the gap with risk-enhancers — family history, Lp(a), hs-CRP, metabolic syndrome, ApoB, ethnicity — that nudge borderline people up. But the decisive tie-breaker in the intermediate zone is to stop estimating and look directly at the arteries with a calcium score or CT angiogram. Risk factors predict; imaging shows.

The Agatston score counts calcified plaque. A score of zero means no calcified plaque and, in most people, a very low 10-year risk — a 'warranty period' of roughly 5–10 years that often justifies safely deferring a statin in an intermediate-risk person. Bands: 1–99 mild, 100–399 moderate, >400 high (in the young, your MESA age/sex percentile matters more than the raw number). A CAC reclassifies about half of intermediate-risk people up or down — for ~$100 and under 1 mSv of radiation. Evidence ●●●●●

A calcium score sees only calcified plaque. Soft, non-calcified, lipid-rich plaque — the vulnerable kind that ruptures and causes many heart attacks, especially in people under 55 — is invisible to it. So a zero is powerfully reassuring in a 65-year-old but can falsely reassure a 40-year-old with early soft plaque or a high Lp(a). It also measures plaque burden, not narrowing: a high score isn't the same as a blockage. This is the exact reason a bare calcium score is not the last word.

The sweet spot is the intermediate-risk 40–70-year-old deciding whether to start a statin. It adds little if you're already high-risk (you're treating regardless) or genuinely low-risk and young. And repeat scans to 'watch it fall' are mostly unhelpful — calcium can rise even as treatment works, because healing soft plaque calcifies and stabilises (the calcium paradox).

Coronary CT angiography (with contrast) images both the channel and the artery wall: total plaque burden, degree of narrowing (graded CAD-RADS), and composition — calcified vs. non-calcified vs. dangerous low-attenuation plaque. In SCOT-HEART, CTA-guided care cut heart attacks by ~41% at 5 years versus standard care. The cost is more radiation and an iodine contrast injection — but it's the definitive non-invasive look at what's in the wall. Evidence ●●●●○

Cleerly (FDA-cleared) runs AI over a CT angiogram to measure every plaque, quantify total and low-attenuation (vulnerable) plaque, and track change over time. The advantage over a calcium score is direct: it detects and quantifies the soft plaque a CAC scores as zero, flags the most rupture-prone type, and gives a per-vessel number you can re-measure to see if treatment is working — turning '0 calcium, all clear' into 'here's your actual soft-plaque burden.'

It's newer and adds cost, and the trials proving that acting on an AI plaque readout changes hard outcomes (rather than simply informing and motivating) are still maturing. It's outstanding for staging disease and for the 'zero calcium but worried' younger patient, and it's a powerful motivator — but it hasn't yet displaced the humble, cheap CAC on proven cost-effectiveness. Evidence ●●●○○ for changing outcomes

Ultrasound of the neck arteries is cheap and radiation-free, and can find plaque a calcium score misses in younger people — a reasonable screen when a CT isn't warranted. The limits: it's operator-dependent, and it looks at the carotids as a proxy, not the coronary arteries themselves.

Statins block HMG-CoA reductase, the liver's cholesterol-making enzyme. The liver responds by pulling more LDL and ApoB out of the blood (it upregulates LDL receptors). They also calm inflammation and help stabilise existing plaque — the 'pleiotropic' effects. Depending on drug and dose they cut LDL ~30–55%; the high-intensity options are atorvastatin 40–80 mg and rosuvastatin 20–40 mg.

The Cholesterol Treatment Trialists' (CTT) meta-analysis of ~170,000 people found that each 1 mmol/L (~39 mg/dL) drop in LDL cuts major vascular events by about 22% per year — proportional, dose-dependent, and consistent across groups. In people who already have established disease or a prior heart attack, the absolute benefit is large and the number-needed-to-treat is excellent. This is the ground essentially no serious expert disputes. Evidence ●●●●●

In people without known heart disease, the relative risk reduction is similar, but the absolute benefit is smaller simply because there are fewer events to prevent. It's meaningful for higher-risk primary prevention and marginal for genuinely low-risk people — and that gap between relative and absolute benefit is exactly what the next section is about. Evidence ●●●●○

Muscle aches are the most-cited complaint — but blinded N-of-1 trials (SAMSON, StatinWISE) found the great majority of perceived statin muscle symptoms are nocebo: people got the same aches on placebo. True pharmacological myopathy is real but uncommon; rhabdomyolysis is very rare. New-onset diabetes is real — roughly one extra case per ~200 person-years at high intensity, dose-related, and mostly in people already prediabetic. Liver-enzyme bumps are usually benign; the 'brain fog' signal hasn't held up in trials. So the risks are real but modest, and the scariest anecdotes are largely nocebo. Evidence ●●●●○

Thoughtful critics — writers around Redberg & Abramson in the BMJ, Robert DuBroff, and others — argue that in low-risk primary prevention the absolute benefit is tiny (some estimates put it at deferring an event by days), that relative-risk headlines ('36% reduction!') oversell a small absolute gain, that much trial data is industry-run with patient-level data not fully open, and that healthy low-risk people get needlessly medicalised. On the narrow point of over-treating the low-risk, this critique has real force. (Where some go too far — denying that LDL causes disease — the evidence is firmly against them; keep those claims separate.)

The counter, from the CTT collaboration and most cardiologists: the causal case for LDL is overwhelming (Mendelian randomisation plus trials line up), the harms are genuinely low and mostly nocebo, and the 'days of life' framing understates benefit because it ignores the non-fatal heart attacks and strokes prevented along the way. Notably, even most defenders now agree that low-risk primary prevention should be a shared decision rather than a blanket mandate — and that a calcium score can personalise it.

Strip away the heat and the facts are agreed: LDL causes plaque (settled), and statins lower it and prevent events (settled). The real disagreement is narrower — the size of the benefit for low-risk people and how honestly it's communicated. That's a values-and-communication question, not a factual one. The way through is your own real risk (ideally with a calcium score), decided with your clinician — not a headline and not a podcast.

Ezetimibe blocks cholesterol absorption in the gut, cutting LDL a further ~15–20% on top of a statin. IMPROVE-IT proved that adding it reduced events — the trial that validated the principle that lowering LDL helps regardless of the mechanism used. It's cheap, generic, and very well tolerated: the obvious first step when a statin alone falls short. Evidence ●●●●○

Bempedoic acid works one step upstream of statins and is activated only in the liver — so muscle tissue never takes it up, making it an option for people who can't tolerate statins. It lowers LDL ~20–25% alone (more with ezetimibe), and CLEAR Outcomes proved it reduced cardiovascular events in statin-intolerant patients. The main watch-outs are a small rise in uric acid and gout risk. Evidence ●●●●○

PCSK9 is a protein that destroys LDL receptors; block it and the liver clears far more LDL. The monoclonal antibodies evolocumab (Repatha) and alirocumab (Praluent), injected every 2–4 weeks, cut LDL a further ~50–60% on top of a statin — down to levels (~30 mg/dL) once thought impossible, with no 'too low' safety floor yet found. FOURIER and ODYSSEY OUTCOMES proved they reduce events; GLAGOV showed actual plaque regression. And uniquely among the mainstays, they lower Lp(a) ~20–30% (statins don't touch it). The safety record is exceptionally clean; the real downsides are the injection and the cost. Evidence ●●●●●

Inclisiran is a small-interfering-RNA (siRNA) that silences PCSK9 production in the liver. It delivers a similar ~50% LDL drop but is dosed just twice a year after a loading dose — a potential game-changer for adherence, since a nurse-given jab every six months is hard to forget. Its cardiovascular outcome trials (ORION-4, VICTORION-2 PREVENT) read out around 2026–27. Evidence ●●●●○ (LDL proven, hard outcomes pending)

For most people PCSK9 inhibitors are added to a statin rather than replacing one — statins are cheap, proven, and oral, while PCSK9s are expensive and injected. But for the truly statin-intolerant, for familial hypercholesterolaemia, and for anyone still far from goal on maximal oral therapy, they are transformative. The 'wins' the hype promises — raw power, clean safety, Lp(a) lowering, and plaque regression — are real; the catch is access and cost, not efficacy.

Elevated Lp(a) affects ~1 in 5 people and independently drives both heart attacks and aortic-valve stenosis — yet none of the standard levers move it. For years the only real options were apheresis (physically filtering it from the blood) in extreme cases, plus driving ApoB and blood pressure harder to compensate. PCSK9 inhibitors and niacin lower it modestly, but neither was designed for the job.

A wave of targeted drugs cut Lp(a) by 65–100%: pelacarsen (antisense; the Lp(a)HORIZON outcome trial reads out ~2025–26), olpasiran (siRNA; OCEAN(a)), lepodisiran, and even an oral option, muvalaplin. The one open question is the big one: does lowering Lp(a) actually cut events? The outcome trials will answer within a couple of years. Evidence ●●●○○ (huge lowering proven; outcome benefit not yet)

Merck's enlicitide (MK-0616) is an oral PCSK9 inhibitor — a once-daily pill delivering a ~50–60% LDL drop, the kind of lowering that until now required an injection. Its Phase 3 programme (CORALreef) reads out around 2025–26. If it lands, it could make PCSK9-level lowering cheap, oral, and mainstream.

Earlier CETP inhibitors failed because they chased HDL. Obicetrapib is different: it's an oral, once-daily drug that meaningfully lowers LDL/ApoB (~50%) and Lp(a) — the axes that actually matter — with Phase 3 outcome data pending. A class left for dead may return by targeting the right thing.

VERVE-101/102 use a single CRISPR base-editing infusion to permanently switch off PCSK9 (or ANGPTL3) in the liver — the prospect of lifelong LDL lowering from one treatment. It's in very early human trials; if it proves safe and durable, it would be transformative. This is the frontier of the frontier — years away, and to be judged on long-term safety.

ANGPTL3 inhibitors (evinacumab is already approved for the rare, severe familial form; siRNA versions like zodasiran target high triglycerides) and better single-pill combinations round out the near-term pipeline. The theme across all of it: get ApoB lower, keep it there, and make doing so easier.

Swap saturated fat for unsaturated: replacing fatty/processed meat, butter, and coconut oil with olive oil, nuts, seeds, and fatty fish lowers LDL/ApoB (a ~10–15% swing is achievable). Viscous fibre: oat β-glucan ≥3 g, psyllium 7–10 g, legumes → ~5–10% LDL. Plant sterols ~2 g → ~7–12%. Cut refined carbs and sugar to lower triglycerides and remnant cholesterol and shift LDL toward the larger, less-atherogenic pattern. These stack: the combined 'Portfolio Diet' reaches ~20–30% LDL reduction. Evidence ●●●●○

Mediterranean eating has the best outcome evidence — PREDIMED found ~30% fewer cardiovascular events with extra-virgin olive oil or nuts. DASH and, broadly, whole-food, high-fibre, plant-forward diets do the same job. Fatty fish helps triglycerides. No single 'superfood' matters as much as the overall pattern you eat most days.

Dietary cholesterol (eggs) matters far less than saturated fat for most people. Low-carb/keto can send LDL and ApoB sharply up in some people ('lean-mass hyper-responders') — so if you eat that way, don't assume it's heart-safe: measure your ApoB. Saturated fat's role is real but more modest than 1980s dogma, and the key is the replacement: swapping it for unsaturated fat helps, swapping it for refined carbs doesn't. And processed meat carries a consistent risk signal. The theme throughout: verify with a blood test rather than trusting the theory.

Visceral fat and insulin resistance drive the whole atherogenic pattern — high triglycerides, low HDL, small dense LDL, high ApoB. Losing visceral fat improves all of them, and even modest loss helps. GLP-1 drugs (semaglutide, tirzepatide) improve the metabolic profile, and SELECT showed semaglutide cut cardiovascular events ~20% in overweight patients without diabetes — a genuine new lever (though only partly via weight). None of this is a licence to skip proven LDL-lowering. Evidence ●●●●○

Aerobic plus resistance training modestly lowers triglycerides and raises HDL, and has a large effect on insulin sensitivity, blood pressure, and plaque stability. Aim for ~150 minutes/week of moderate activity, ideally with both zone-2 endurance and something that builds VO₂ max. It doesn't lower LDL much directly — its value is metabolic and vascular, protecting the artery wall rather than moving the LDL number.

Smoking is one of the most potent accelerators of plaque and rupture — quitting is among the highest-yield things a person can do for their arteries, full stop. Alcohol raises triglycerides, and the old 'red wine is heart-healthy' story has largely collapsed — there's no cardiac threshold worth chasing. Poor sleep and sleep apnoea worsen blood pressure, insulin resistance, and inflammation; chronic stress feeds the same fire via cortisol and inflammation. These move plaque biology directly.

Red yeast rice works because it is low-dose lovastatin (~15–25% LDL) — but with an unregulated, wildly variable dose and possible citrinin contamination, so a prescription statin is the safer version of the same thing. Bergamot (~8–20%) and berberine (~15–20%, and a mild natural PCSK9 nudge; GI upset; avoid in pregnancy) have decent surrogate evidence. Plant sterols (7–12%) and viscous fibre/psyllium (5–10%) are the safest, food-based options; pantethine ~10–15%. All lower a number; none has a hard-outcome trial. Evidence ●●●○○ (surrogate only)

Omega-3 does not lower LDL (high doses can nudge it up); it lowers triglycerides ~20–30%. Only prescription EPA (icosapent ethyl, 4 g) has outcome data — REDUCE-IT cut events ~25%, though that's caveated by the mineral-oil-placebo controversy. Ordinary OTC fish oil is largely null for events, and 4 g/day raises atrial-fibrillation risk. Use it for high triglycerides, not to move ApoB. Evidence ●●●●○ (TG lowering) / ●●○○○ (OTC outcomes)

Nattokinase is a potent fibrinolytic enzyme: unlike statins (which cut how much cholesterol the liver makes), it breaks down fibrin — the sticky protein scaffold that holds soft plaque and clots together — which may let the body clear buildup over time. It's genuinely active in humans. The single most useful thing to know is that dose decides everything: the ~2,000 FU in most shop-shelf supplements does nothing for cholesterol or plaque, and even 3,600 FU/day showed no effect — the benefits only appeared at ~6,000–10,800 FU/day, so most people take far too little to matter.

The largest is a 2022 trial of 1,062 people with mild atherosclerosis and high cholesterol. Over 12 months, the high-dose arm (10,800 FU/day) saw a ~36.6% average reduction in carotid plaque, plus lower LDL and triglycerides and higher HDL — while the low-dose arm (3,600 FU) changed nothing (Frontiers in Cardiovascular Medicine, 2022). An earlier head-to-head (Ren 2017, ~76 patients, 26 weeks) reported nattokinase shrank plaque more than simvastatin (~37% vs ~12%) and raised HDL, though the statin lowered LDL better.

Genuinely intriguing, and the dose-response is a point in its favour. But temper it: these standout results come largely from a single research orbit, aren't independently replicated, and the strongest-looking designs lack a true placebo (Ren compared against a statin, not placebo) — while a separate placebo-controlled 3-year trial (NAPS) found no effect on plaque progression. So it's promising and worth watching, not the settled, replicated evidence a statin has. Evidence ●●○○○ — dose-dependent and intriguing, but unreplicated

At the therapeutic doses that actually move plaque (6,000–10,000+ FU), nattokinase acts as a real systemic blood thinner, with documented bleeding events (including brain bleeds and a fatal case). Talk to your doctor before starting — especially if you take aspirin or an anticoagulant, or have any surgery coming up — and never substitute it for prescribed anticoagulation.

Niacin moves every lipid number the 'right' way, yet AIM-HIGH and HPS2-THRIVE showed no benefit and net harm on a statin background — the textbook lesson that a better panel isn't a better outcome. CoQ10 doesn't lower cholesterol; its evidence for easing statin muscle aches is mixed but it's safe and worth trying before abandoning a statin. Garlic, green tea, and soy give tiny, inconsistent LDL effects. Vitamin K2 has a plausible anti-calcification hypothesis and one supportive imaging trial but no outcome data (and it antagonises warfarin). Evidence ●●○○○ or weaker

There is no peptide with credible human evidence for lowering cholesterol or reversing plaque. ApoA-I / HDL-mimetic peptides — the most-studied class — have failed in human trials; BPC-157's cardiovascular claims are extrapolated from rodents; and injectable 'cardiac peptides' sold online are unregulated and potentially unsafe (sterility, dosing, contamination). It's an active research frontier with zero validated products today. Be candid with yourself: here, the marketing is far ahead of the science. Evidence ○○○○○ in humans

Every load-bearing number comes from a named guideline, trial, meta-analysis or regulatory document, with the evidence dots (●●●●● = strong and replicated, down to ○○○○○ = essentially unproven) reflecting study design, size and replication. Figures are group averages; individuals vary widely. Doses and thresholds are typical adult figures from the literature, not instructions.

Plaque biology & regression: the response-to-retention model and the IVUS regression/stabilisation trials (REVERSAL, ASTEROID, SATURN, GLAGOV, HUYGENS, PACMAN-AMI). Causality: Mendelian-randomisation studies, the CTT meta-analyses, and the EAS ApoB/LDL causal consensus. Tests & targets: the ApoB and Lp(a) consensus statements and the lipid-guideline target tables. Imaging: MESA (calcium percentiles and the power of zero), SCOT-HEART (CT angiography), and Cleerly's validation literature. Statins — both sides: the CTT collaboration for benefit, SAMSON and StatinWISE for the nocebo effect, and the BMJ overtreatment critiques. Non-statins: IMPROVE-IT (ezetimibe), CLEAR Outcomes (bempedoic acid), FOURIER/ODYSSEY/GLAGOV (PCSK9), and the ORION programme (inclisiran). Lp(a) & pipeline: the pelacarsen and olpasiran trials, obicetrapib, oral PCSK9 (enlicitide), and the VERVE gene-editing programmes. Lifestyle: PREDIMED, the Portfolio Diet, and SELECT (semaglutide). Supplements: the NCCIH natural-products reviews, REDUCE-IT vs. STRENGTH, HPS2-THRIVE (niacin), and the nattokinase trials (Ren 2017 vs. the placebo-controlled NAPS).