Lipid management
What ages your arteries isn't one cholesterol reading — it's the lifetime dose of cholesterol-carrying particles they've been exposed to, and the count of those particles predicts cardiovascular risk better than LDL cholesterol does. In a UK Biobank study of 375,544 adults, about one in five had a particle count that clearly disagreed with their LDL cholesterol, and in half of them it was higher than the cholesterol number suggested.Du et al., Eur J Prev Cardiol 2025 And most statin muscle complaints are not caused by the drug: in blinded trials, more than 90% of the aches reported by people on a statin were not due to the statin.Cholesterol Treatment Trialists' Collaboration, Lancet 2022
The modern frame for cardiovascular prevention has shifted from "fix the cholesterol number when it goes up in midlife" to "reduce the cumulative lifetime burden of cholesterol-carrying particles in the arterial wall, the earlier the better." Genetic and biomarker evidence both point in the same direction: the dose × time integral of apolipoprotein B (apoB)–containing particle exposure determines plaque progression, and intervening earlier produces vastly more benefit per unit of cholesterol lowered than intervening late. The 2026 ACC / AHA Guideline on the Management of Dyslipidemia codifies the shift — reinstating numeric LDL-C and non-HDL-C goals, using coronary artery calcium scoring selectively when the treatment decision is uncertain, recommending a lipoprotein(a) measurement for every adult, and laying out escalation from statin to ezetimibe, bempedoic acid and PCSK9 (proprotein convertase subtilisin/kexin type 9) inhibition. For the genetically driven Lp(a) burden that about 20% of adults carry, there is still no proven treatment: the first outcomes trial of an Lp(a)-lowering drug missed its primary endpoint in 2026. This article walks the full picture.
The shift: cumulative exposure, not snapshot cholesterol
Evidence: Strong — converging Mendelian-randomization, genetic, and long-cohort (CARDIA) data.
The historical model of cholesterol-driven heart disease treated a lipid panel as a current state — high LDL today, treat today. The modern biological view, codified by Ference, Braunwald, and Catapano in a 2024 Nature Reviews Cardiology synthesis, is that atherosclerosis is the integral of apoB-containing lipoprotein exposure over time, not a snapshot measurement.[1]
Each apoB-containing particle that traverses the arterial wall has a probability of being trapped and retained. Over decades, retained particles oxidise, recruit macrophages, drive plaque formation, and progressively narrow vessels. The total disease burden is determined by how high your particle count has been multiplied by how long it has been that high.
The CARDIA study — 4,958 asymptomatic adults enrolled at age 18–30 and followed for a median 16 years after they turned 40 — quantified this directly. Cardiovascular event risk tracked the area under each person's curve of LDL cholesterol (LDL-C) plotted against age — each additional 100 mg/dL-years of cumulative exposure raised risk by about 5% (hazard ratio 1.053), independent of their LDL-C at any single point in time. Crucially, the timing of that exposure mattered independently — the same area accumulated at a younger age produced a larger risk increase than the same area accumulated later.[2]
Mendelian-randomization studies — using genetic variants that produce naturally low lifelong LDL as a proxy for lifelong pharmacological lowering — show up to a threefold greater proportional risk reduction per unit of LDL lowered when the lowering begins early, versus the same reduction starting in midlife with statins. Atherosclerosis you never grow is much easier to prevent than atherosclerosis you've already accumulated.[3] The longevity implication: lipid management isn't a problem to solve at age 55; it's a cumulative exposure to defend against starting much earlier.
The biomarker that matters: apoB, not LDL cholesterol
Evidence: Strong (apoB is at least as good a predictor as LDL-C, and superior under discordance); Moderate for the incremental outcome benefit of acting on apoB rather than LDL-C, which rests on observational cohorts.
For most of medical history, the primary cholesterol metric has been low-density lipoprotein cholesterol (LDL-C) — a measurement of how much cholesterol mass is being carried inside LDL particles. The clean biology turns out to be one level deeper.
Every atherogenic particle in your bloodstream — LDL, intermediate-density lipoprotein (IDL), very-low-density lipoprotein (VLDL), and lipoprotein(a) — carries exactly one molecule of apolipoprotein B (apoB) on its surface. Measuring plasma apoB therefore gives a direct stoichiometric count of how many atherogenic particles you have in circulation. LDL-C, by contrast, measures only the cholesterol mass inside one subset of those particles, and the cholesterol-per-particle ratio varies considerably between individuals.[4]
When particles are smaller, denser, and depleted of cholesterol — as in metabolic syndrome, insulin resistance, and some genetic patterns — the LDL-C number can look reassuring while the particle count is dangerously high.
The discordance is bigger than the field assumed
A 2024 JAMA Cardiology analysis by Sayed, Sniderman and colleagues quantified the LDL-C / apoB discordance directly in 12,688 adults from the US National Health and Nutrition Examination Survey (NHANES).[5] At a standard LDL-C of 100 mg/dL, the median apoB was 80 mg/dL. The 95% range, however, spanned 66 to 99 mg/dL of apoB at that same LDL-C. Higher-than-expected apoB was more common in people with metabolic risk factors, but it wasn't restricted to them: apoB varied widely at any given LDL-C even in metabolically healthy adults (normal weight, triglycerides under 150 mg/dL, no diabetes). The variation was smallest around non-HDL cholesterol, which tracks apoB more closely than LDL-C does.
A 375,544-person UK Biobank analysis of discordance confirmed the clinical consequences. In adults whose apoB was clearly higher than their LDL-C predicted, the risk of major cardiovascular events ran about 6–11% higher, depending on how fully the analysis adjusted for other factors (6% after accounting for triglycerides, HDL cholesterol and Lp(a)), independently of absolute LDL-C levels.[6] In separate mediation analyses, VLDL particles and triglycerides each explained about a quarter of the excess risk (25–27%), and the authors conclude that the extra risk was driven mainly by the number of apoB-carrying particles.
The 2026 ACC / AHA dyslipidemia guideline gives apoB a supporting role. For people on lipid-lowering treatment, measuring it is "reasonable" once their LDL-C and non-HDL-C goals are met, to decide whether to intensify — especially with cardiovascular disease, cardiovascular-kidney-metabolic syndrome, type 2 diabetes or high triglycerides; before treatment, it "may be reasonable" to refine risk.[7] The guideline's own summary singles out triglycerides above 200 mg/dL, diabetes and an already low LDL-C (below 70 mg/dL) as the situations where it helps most.[8] For longevity-oriented adults running a baseline workup, a one-time apoB measurement is probably worth more than the LDL-C value on the same panel.
| Biomarker | What it measures | When it matters most |
|---|---|---|
| LDL-C | Cholesterol mass inside LDL particles | Standard target metric; ubiquitous, cheap, well-understood |
| Non-HDL-C | Total atherogenic cholesterol mass | Useful when triglycerides are high; doesn't quantify particle number |
| ApoB | Direct count of atherogenic particles | The definitive measure of true particle burden when LDL-C and apoB disagree |
Risk estimation: PREVENT, PCE, and the calcium score
Evidence: Moderate for the risk calculators (they rank people well, but their absolute estimates can drift, possibly most in people who never start a statin); Strong for coronary artery calcium as a predictor of risk; Weak for CAC-guided treatment improving outcomes — in randomised trials so far, scanning modestly improved some risk factors, mainly cholesterol, and may have helped statin adherence, but none has shown fewer heart attacks or deaths.[9]
Until recently, the standard US cardiovascular risk calculator was the Pooled Cohort Equations (PCE) — built on older datasets, using race as a biological proxy, and with no kidney function. In 2023 the American Heart Association published its successor, the PREVENT equations (Predicting Risk of cardiovascular disease EVENTs),[10] and the 2025 hypertension and 2026 dyslipidemia guidelines now recommend them in place of the PCE.[11][12]
PREVENT adds estimated glomerular filtration rate and statin use, and can optionally include HbA1c, urine albumin-to-creatinine ratio and a Social Deprivation Index; body mass index enters only its heart-failure and death models, not the atherosclerotic-disease equation used for statin decisions.[13] It estimates both 10-year and 30-year risk and removes race as a predictor.
One caveat worth knowing. PREVENT gives lower risk estimates than the older equations, and it may understate the risk of someone who never starts treatment. In a study of 193,885 adults aged 40–75 without diabetes or cardiovascular disease in one California health system, PREVENT matched observed 10-year risk across the whole group. But among people who never started a statin during follow-up, observed risk ran higher than predicted: 8.2% of those with a predicted risk of 5% to <7.5% had an event, against an average prediction of 6.1%, and the older Pooled Cohort Equations came closer in that group.[14] Who started a statin was not randomised, so part of the gap may reflect who chose treatment. Larger validations, including one across 30 US health systems by PREVENT's developers, found it well calibrated overall, though without looking separately at untreated people.[15] The practical lesson: treat a borderline PREVENT score as a starting point, not a verdict, especially with a strong family history or long-standing high LDL-C.
Coronary artery calcium: turning probability into anatomy
For adults at intermediate (5% to <10%) or selected borderline (3% to <5%) 10-year PREVENT risk whose treatment decision is still uncertain, the 2026 guideline recommends coronary artery calcium (CAC) scoring — a non-contrast computed-tomography measurement — as a selective tool, in men from age 40 and women from 45.[16] The principle: probability becomes anatomy. If you have measurable coronary calcium, you have subclinical atherosclerosis; you're no longer in a statistical grey zone.
| CAC score (Agatston units) | What it means | 2026 US guideline LDL-C goal |
|---|---|---|
| 0 | No calcified plaque; low near-term risk, but the "warranty" lasts only about 3–7 years | No goal; deferring a statin and rescanning in 3–7 years is reasonable only in the situation described below |
| 1–99, and below the 75th percentile for age, sex and race | Definite but mild subclinical disease | <100 mg/dL (2.6 mmol/L) |
| 100–299, or at or above the 75th percentile with a score below 300 | Moderate plaque burden | <70 mg/dL (1.8 mmol/L) |
| 300–999 | Severe plaque burden | <70 mg/dL (1.8 mmol/L); intensifying to <55 mg/dL (1.4 mmol/L) is reasonable |
| ≥1,000 | Extensive plaque burden | <55 mg/dL (1.4 mmol/L) |
These goals follow the guideline's recommendations for subclinical atherosclerosis.[17]
A CAC of zero buys time, but less than is often claimed. In the Multi-Ethnic Study of Atherosclerosis (MESA), about half of people who started with a zero score had some calcium within 10 years; the estimated "warranty period" ran about 3 to 7 years depending on sex and ethnicity, and diabetes shortened it by 35–38%, to about 3 years in men and 4 in women.[18] The 2026 guideline therefore treats deferral as reasonable only for adults at intermediate (or selected borderline) risk who prefer to avoid medication and who have none of these higher-risk conditions: familial hypercholesterolemia or an LDL-C of 190 mg/dL (4.9 mmol/L) or more, diabetes after age 40, current smoking, or a strong family history of premature heart disease. Even then, it advises a repeat scan in 3 to 7 years. Any detectable calcium is a different situation: the guideline recommends starting lipid-lowering treatment, particularly at a score of 100 or more or at or above the 75th percentile.[19]
Statins in primary prevention: what the trials actually showed
Evidence: Strong — multiple large randomised trials with hard cardiovascular endpoints.
The bedrock of pharmacological lipid management remains statins — competitive inhibitors of the rate-limiting enzyme in hepatic cholesterol synthesis (3-hydroxy-3-methylglutaryl coenzyme A reductase, HMG-CoA reductase). The drop in intracellular cholesterol forces hepatocytes to upregulate LDL receptors, which then clear circulating apoB-containing particles. Three primary-prevention trials anchor the modern understanding.
JUPITER — 17,802 apparently healthy adults with normal LDL-C (median 108 mg/dL) but elevated high-sensitivity C-reactive protein (≥2.0 mg/L, a marker of low-grade inflammation) — was stopped early at 1.9 years for an extraordinary 44% reduction in the primary composite of cardiovascular events on rosuvastatin 20 mg.[20] The number needed to treat was 95 over 2 years. The trial also recorded more physician-reported type 2 diabetes on rosuvastatin — 270 cases against 216, about 25% more (HR 1.25), with the excess confined to people who already had a diabetes risk factor.[21] Pooled blinded-trial data show this is a dose-dependent class effect, not a high-intensity quirk: compared with placebo, low- or moderate-intensity statins raised new diabetes diagnoses by about 10% (1.3% vs 1.2% a year) and high-intensity statins by about 36% (4.8% vs 3.5% a year in those trials). About 62% of the extra cases were in people whose blood sugar was already in the top quarter at baseline, and any cardiovascular cost is already counted in the trials' overall event reduction.[22] It belongs in the consent conversation.
HOPE-3 — a more generalisable cohort of 12,705 intermediate-risk adults — tested low-dose rosuvastatin 10 mg over 5.6 years and produced a 24% reduction in the primary composite of cardiovascular death, non-fatal heart attack, and stroke (about 1 event prevented for every 91 people treated over a median 5.6 years).[23] This lower dose produced no excess of diabetes in the trial, though it did produce slightly more muscle symptoms (5.8% vs 4.7%) and cataract surgery (3.8% vs 3.1%), and a substudy of participants aged 70 and over found no effect on cognition.[24]
MEGA — 7,832 Japanese adults with hypercholesterolaemia and no prior coronary disease or stroke, randomised to diet alone versus diet plus low-dose pravastatin (10–20 mg) — produced a 33% reduction in the primary endpoint of coronary heart disease (66 vs 101 events; hazard ratio 0.67, 95% confidence interval 0.49–0.91) over a mean 5.3 years, despite the lower-baseline-risk population and a modest 18% LDL-C reduction. (A confidence interval is the range in which the true effect most plausibly lies; when it sits entirely below 1.0, as here, the benefit is unlikely to be a chance finding.)[25] The safety profile was clean: no excess of malignancy or other serious adverse events.
A practical observation across these and other primary-prevention trials: the absolute benefit builds steadily with time. In a meta-analysis of eight trials in adults aged 50 to 75, about 0.3 major events were prevented per 100 people treated at one year, 1.3 at three years and 2.5 at five years; only one of the eight trials found fewer deaths from any cause.[26] Statins aren't a "see if it works" intervention: the payoff depends on staying on them. For people over 70, a large 2026 trial now speaks directly: in 9,971 adults aged 70 or older without cardiovascular disease, diabetes or dementia, atorvastatin 40 mg cut major cardiovascular events by about 30% (10.9 vs 15.5 events per 1,000 person-years over a median 5.9 years) but did not lengthen survival free of dementia or persistent disability.[27]
The SAMSON trial: most statin "side effects" are not caused by the drug
Evidence: Moderate — one rigorous n-of-1 crossover trial (n=60), converging with the ASCOT-LLA blinded-vs-unblinded pattern and blinded RCT symptom rates.
The dominant reason statins are discontinued is muscle symptoms — myalgia, weakness. Real-world surveillance puts the reported rate above 10%. In blinded trials, almost all of those complaints turn out not to be caused by the drug: across 19 double-blind trials of 123,940 people followed for about four years, 27.1% of those on a statin and 26.6% of those on placebo reported muscle pain or weakness — a 3% relative excess (rate ratio 1.03), concentrated in the first year and somewhat larger on high-intensity regimens. More than 90% of the muscle complaints reported by people taking a statin were not caused by it.[28] The ASCOT-LLA trial showed the same thing from another angle: muscle complaints were reported at the same rate on atorvastatin and placebo while patients were blinded (about 2% a year in each arm), but about 40% more often among statin users once the trial became open-label.[29]
The SAMSON trial (Self-Assessment Method for Statin Side-effects Or Nocebo) was designed to settle the question.[30] Sixty adults who had stopped statins, with no intention of restarting, because of intolerable side effects that appeared within two weeks of starting — muscle aches in most of them — were enrolled in an n-of-1 crossover design. They received 12 monthly bottles in randomised order: 4 containing atorvastatin 20 mg, 4 containing placebo, and 4 empty. Daily symptom intensity was tracked via smartphone on a 1–100 scale.
The results landed cleanly:
- No-tablet months: mean symptom score 8.0.
- Statin months: mean symptom score 16.3.
- Placebo months: mean symptom score 15.4 — statistically indistinguishable from the statin months (p = 0.388).
The calculated "nocebo ratio" — the share of the extra symptoms on statin months (above no-tablet months) that placebo tablets also produced — was 0.90. About 90% of the symptom burden that seemed to come from the statin in these people, who had already quit statins, was produced by the expectation of harm rather than the pharmacology. Six months after the trial, half of the 60 were taking a statin again.
The implication for practice isn't that statin myopathy is fake — true rhabdomyolysis with elevated creatine kinase is a documented if rare event. But the typical myalgia that leads patients to abandon therapy is mostly not caused by the drug, and running and discussing a personal trial got half of these patients back on treatment. Beyond muscle effects and diabetes, blinded trials confirm only a few small harms: in 19 double-blind trials of 123,940 people, abnormal liver-enzyme results were about 41% more common on a statin (0.30% vs 0.22% of participants a year), with slight excesses of urine-test changes and fluid swelling (oedema), but no excess of memory problems, depression, sleep disturbance, nerve damage or sexual dysfunction, despite their appearance on product labels.[31]
CoQ10 is widely tried for statin muscle symptoms — statins do lower it — but the randomized evidence is mixed, and against a 90% nocebo background any uncontrolled "response" is hard to attribute to the supplement. It is cheap and safe enough to trial, but not a substitute for the n-of-1 frame above. See CoQ10.
When statins aren't enough: ezetimibe, bempedoic acid, PCSK9 inhibitors
Evidence: Strong — each agent below has a dedicated cardiovascular-outcomes trial (IMPROVE-IT, CLEAR Outcomes, FOURIER, ODYSSEY Outcomes, and VESALIUS-CV before a first event).
The 2010s saw the rise of a clean clinical principle: lowering apoB-carrying particles works, whichever drug does it. The historical idea that statins' benefit came from non-cholesterol "pleiotropic" effects has been largely settled — drugs that clear apoB particles by different routes produce benefit in proportion to how far they lower them. Not every cholesterol-lowering drug qualifies, though: the CETP inhibitor evacetrapib cut LDL cholesterol by about 31% with no reduction in events,[32] and genetic evidence suggests benefit tracks the fall in apoB particles rather than in LDL cholesterol as such.[33]
Ezetimibe (IMPROVE-IT)
Ezetimibe blocks intestinal cholesterol absorption via the Niemann-Pick C1-Like 1 transporter. The IMPROVE-IT trial added ezetimibe to simvastatin in 18,144 patients after acute coronary syndrome and lowered LDL-C from about 70 to 54 mg/dL (1.8 to 1.4 mmol/L). Over 7 years, the main composite of cardiovascular events fell from 34.7% to 32.7% — about 6% lower risk (HR 0.936), or 2 fewer events per 100 patients.[34] Muscle, gallbladder and liver side effects and cancer were no more common. IMPROVE-IT showed that adding a non-statin LDL-lowering drug to a statin further improves outcomes.
Bempedoic acid (CLEAR Outcomes)
Bempedoic acid is an inhibitor of ATP-citrate lyase, an enzyme upstream of HMG-CoA reductase in the cholesterol synthesis pathway. Crucially, it's a prodrug activated by the enzyme very-long-chain acyl-CoA synthetase 1 (ACSVL1), which is found primarily in the liver,[35] so the drug should act on hepatic cholesterol synthesis while largely sparing muscle. The design targets the muscle-symptom problem; in CLEAR Outcomes, muscle aches were no more common than on placebo.[36]
The CLEAR Outcomes trial randomised 13,970 adults who were unable or unwilling to take statins because of side effects, and who had elevated LDL-C (mean 139 mg/dL), to bempedoic acid or placebo.[37] Headline results published in NEJM 2023:
- 13% reduction in the primary four-component composite of major adverse cardiovascular events (MACE — cardiovascular death, heart attack, stroke, or coronary revascularisation; hazard ratio 0.87).
- 15% reduction in the three-component version (cardiovascular death, heart attack, stroke; hazard ratio 0.85), and a 23% reduction in heart attacks specifically.
- No significant effect on stroke, cardiovascular death or death from any cause.
- No new-onset diabetes signal (unlike statins): 11.1% vs 11.5% of participants without diabetes developed it.[38]
- Fewer major limb events in a secondary analysis of the 1,624 participants with peripheral artery disease (about 36% fewer first events, HR 0.64).[39]
For an adult who cannot take a statin because of side effects, bempedoic acid is now a guideline-backed alternative, and in high-risk patients it can be added to a statin when LDL cholesterol is still above goal.[40] It is not a substitute for a statin you can tolerate. On its own it lowers LDL cholesterol by only about 21–24% (17–18% on top of a statin), against 50% or more for a high-intensity statin,[41] and in CLEAR Outcomes it did not significantly reduce cardiovascular death. It also has side effects of its own: compared with placebo, more people developed gout (3.1% vs 2.1%) and gallstones (2.2% vs 1.2%), and uric acid, creatinine and liver enzymes rose slightly,[42] so the European guideline advises caution in anyone with a history of gout.[43]
PCSK9 inhibitors (FOURIER, ODYSSEY Outcomes)
The discovery of the PCSK9 protein reset the upper limit of pharmacological cholesterol lowering. PCSK9 binds hepatic LDL receptors and tags them for destruction, limiting the cell's clearing capacity. Monoclonal antibodies that bind and neutralise circulating PCSK9 prevent this — allowing massive sustained upregulation of LDL receptors and rapid clearance of circulating apoB particles.
FOURIER added evolocumab to maximally-tolerated statins in 27,564 adults with established cardiovascular disease and produced:[44]
- 59% reduction in LDL-C versus placebo — median fell from a baseline of 92 mg/dL to 30 mg/dL.
- 15% reduction in the primary 5-component MACE composite.
- 20% reduction in the 3-component MACE composite.
The event-curve separation widened over time, exactly as cumulative-exposure theory predicts — early lowering produces small acute benefit; sustained lowering produces compounding benefit over years. ODYSSEY Outcomes replicated the pattern with alirocumab in 18,924 post-acute-coronary-syndrome patients: about 15% fewer major events (HR 0.85).[45] PCSK9 antibodies now also work before a first event. In VESALIUS-CV, 12,257 adults with atherosclerosis or high-risk diabetes but no previous heart attack or stroke, and an LDL-C of at least 90 mg/dL (2.3 mmol/L), received evolocumab or placebo. Over a median 4.6 years, evolocumab cut coronary death, heart attack or ischaemic stroke by about a quarter (HR 0.75) — by five years, 6.2% of patients versus 8.0% on placebo — with no excess of safety events.[46]
Two practical lessons from the PCSK9 trials. First, very low LDL-C has so far looked safe: in the open-label extension of FOURIER, adults with established cardiovascular disease reached a median LDL-C of 30 mg/dL (0.8 mmol/L) within 12 weeks and were followed for a median 5 years (up to 8.4 years on the drug), and serious adverse events, new diabetes, haemorrhagic stroke and memory or thinking problems were no more common than on placebo in the original trial, nor did they rise over time. Starting early also paid off: those first given evolocumab in the original trial had about 20% fewer cardiovascular deaths, heart attacks and strokes than those who started about two years later (HR 0.80).[47] Second, for many people with established disease, the limit on how low LDL-C goes is cost and access, not what the drugs can do.
Lipoprotein(a): the silent monogenic risk
Evidence: Strong for Lp(a) as a causal, genetically-fixed risk factor; Weak for treatment benefit — the first outcomes trial of an Lp(a)-lowering drug (pelacarsen) missed its primary endpoint in 2026, and the remaining trials are not expected to finish before 2028.
For roughly 20% of people — about 1.4 billion worldwide — an inherited particle that statins do not lower adds substantially to cardiovascular risk: lipoprotein(a), or Lp(a).[48] At 125 nmol/L (50 mg/dL), a level about one person in five reaches, risk is about 1.4 times that of someone with a typical level; at 250 nmol/L (100 mg/dL) it is about double.[49]
Lp(a) is an LDL-like particle with one bonus feature: a covalently attached apolipoprotein(a) tail that's structurally similar to plasminogen. That makes Lp(a) atherogenic (via the apoB core driving plaque) and pro-inflammatory (partly through the oxidised phospholipids it carries). Because apo(a) resembles plasminogen, Lp(a) has long been suspected of promoting clots as well, but the European Atherosclerosis Society consensus judges that role unproven in people. The combination produces excess risk for both coronary disease and calcific aortic valve stenosis: in people aged 45–54, aortic stenosis was about three times as common when Lp(a) was above the 80th percentile.[50]
Lp(a) levels are >90% genetically determined. Adult levels are usually reached by about age 5 and then stay fairly stable, although they tend to rise in women at menopause. Lifestyle barely moves them: dietary changes shift Lp(a) by about 10–15% at most, in either direction, and exercise has little or no effect.[51] Statins do not lower Lp(a) and may raise it modestly in some people, but the average rise is small — about 1 mg/dL compared with placebo — and is no reason to stop a statin.[52] Ezetimibe and bempedoic acid have no effect.[53] PCSK9 antibodies lower Lp(a) by roughly 15–30%, and lipoprotein apheresis also lowers it but is resource-intensive.[54] In the FOURIER trial of people with established cardiovascular disease, evolocumab cut coronary events by about 23% (HR 0.77) when baseline Lp(a) was above the median, against a non-significant 7% (HR 0.93) when it was lower — 2.49 versus 0.95 fewer events per 100 patients over 3 years.[55] Because that is a subgroup finding, it cannot show whether the extra benefit came from lowering Lp(a) or LDL cholesterol.
The 2026 ACC / AHA guidelines now explicitly recommend that every adult have Lp(a) measured at least once in their lifetime. The test is cheap, the result is stable, and it identifies adults whose cardiovascular risk profile is fundamentally different from what their LDL-C and apoB suggest.
Targeted Lp(a)-lowering drugs
A class of drugs designed specifically to block Lp(a) production — most by silencing the gene's messenger RNA in the liver, one by stopping the particle from forming — has been or is being tested in phase 3 outcomes trials:
| Drug | Type | Maker | Schedule | Lp(a) reduction (phase 2) | Outcomes trial |
|---|---|---|---|---|---|
| Pelacarsen | Antisense oligonucleotide | Novartis / Ionis | Monthly subcutaneous | About 72% at a monthly dose (80% at the most intensive weekly dose)[56] | Lp(a) HORIZON: missed its primary endpoint (topline, September 2026) |
| Olpasiran | Small interfering RNA (siRNA) | Amgen | Every 12 weeks subcutaneous | >95% at doses of 75 mg and above[57] | OCEAN(a)-Outcomes; estimated primary completion 2028[58] |
| Lepodisiran | Long-acting siRNA | Eli Lilly | Once or twice yearly | About 94%, averaged over a year at the highest dose[59] | ACCLAIM-Lp(a); estimated primary completion 2029[60] |
| Muvalaplin | Oral small molecule (blocks Lp(a) formation) | Eli Lilly | Daily oral | Up to about 86% (intact-Lp(a) assay) or about 70% (apo(a)-based assay) at 12 weeks[61] | MOVE-Lp(a); estimated primary completion 2031[62] |
The Lp(a) HORIZON trial gave pelacarsen (80 mg injected under the skin monthly) or placebo to 8,323 people with established cardiovascular disease and Lp(a) of at least 70 mg/dL, on top of their usual lipid-lowering treatment. It was the first cardiovascular-outcomes trial of any Lp(a)-lowering drug, and it missed its primary endpoint, which was tested both in everyone enrolled and in the prespecified group with Lp(a) of 90 mg/dL or more: topline results announced in September 2026 showed that pelacarsen lowered Lp(a) but did not significantly reduce the combined rate of cardiovascular death, heart attack, stroke and urgent coronary revascularisation.[63] No hazard ratio or event rates have been released; the full data are to be presented at a medical congress. So no trial has yet shown that lowering Lp(a) prevents heart attacks or strokes. Lp(a) remains an independent risk factor and is still worth measuring once, because a high value is a reason to control everything else harder.[64] Whether other Lp(a)-lowering drugs can reduce events is now the question for the trials in the table above, none of which is expected to finish before 2028.
What the 2026 dyslipidemia guidelines actually say
Evidence: Strong — current US multisociety guideline consensus, built on the trials above.
The 2026 ACC / AHA / Multisociety guideline replaced the 2018 cholesterol guideline with a more aggressive, goal-based framework built on LDL-C and non-HDL cholesterol (the cholesterol carried by all atherogenic particles), with universal Lp(a) testing and a supporting role for apoB.[65] The headline changes:
- Numeric LDL-C and non-HDL-C goals are back. After years of "high-intensity statin" rather than specific numbers, the guideline reinstated explicit goals, used alongside the percentage fall in LDL-C to decide when to intensify treatment:
- <100 mg/dL (2.6 mmol/L) for borderline (3% to <5% 10-year PREVENT risk) and intermediate risk (5% to <10%), once a statin has been started
- <70 mg/dL (1.8 mmol/L) for high-risk primary prevention (10% or more), and for established cardiovascular disease not at very high risk
- <55 mg/dL (1.4 mmol/L) for established cardiovascular disease at very high risk (two or more major events, or one plus at least two high-risk conditions) or with chronic kidney disease (stage 3 or worse) together with established cardiovascular disease — which, by the guideline's own summary, will cover most people who have had a heart attack or stroke; it is an optional goal for other people with cardiovascular disease
- PREVENT replaces the Pooled Cohort Equations — with the caveat that, in at least one large cohort, it underestimated risk in people who never started a statin.
- Coronary artery calcium scoring is used selectively — in men from 40 and women from 45 at intermediate (or selected borderline) risk, when the decision to treat remains uncertain.
- ApoB measurement is reasonable once LDL-C and non-HDL-C goals are met on treatment, especially with cardiovascular disease, cardiovascular-kidney-metabolic syndrome, type 2 diabetes or high triglycerides, and may be reasonable before treatment to refine risk.
- Universal Lp(a) screening at least once in a lifetime.
- Earlier screening for familial hypercholesterolemia (FH). A lipid profile is recommended for every child at 9–11 years, and from age 2 in children with a close relative who has FH, very high cholesterol or premature heart disease. Adults should be screened from age 19 and at least every 5 years. In young adults, drug treatment should be considered early when LDL-C is 160 mg/dL (4.1 mmol/L) or higher, or when there is a strong family history of premature heart disease.
- Statin first, then add what is needed. On top of the highest tolerated statin, the guideline lets clinicians add ezetimibe, a PCSK9 antibody or bempedoic acid, chosen by how much further LDL-C must fall and by patient preference, rather than in a fixed order; for very-high-risk patients, ezetimibe and/or a PCSK9 antibody should be added. Inclisiran, an injection that also blocks PCSK9 (given every 6 months after two starting doses), is a second-line option because its outcome trials have not yet reported.[66]
Practical guidance
- Get one baseline apoB measurement, and one lifetime Lp(a) measurement. Both are inexpensive, both add genuine information that an ordinary lipid panel doesn't capture. Most adults have never had either drawn. Both belong on the broader midlife lab panel, alongside the metabolic and inflammatory markers that interact with apoB-driven risk.
- Use PREVENT as a starting point, not a verdict. In one large study it underestimated risk in people who never started a statin. Don't let a borderline 10-year score (3% to <5%) talk you out of treatment if your family history, lifelong LDL trajectory or other risk-enhancing factors suggest higher real risk; at an intermediate score (5% to <10%), the guideline generally recommends a statin anyway.
- For intermediate (or selected borderline) calculated risk, if you are unsure about starting a statin, consider a coronary artery calcium scan. A zero score buys a few years — the guideline advises rescanning in 3 to 7 — but only if you have no higher-risk condition such as diabetes after 40, current smoking or a strong family history. Any detectable calcium moves you from probability to anatomy: the LDL-C goal is <100 mg/dL for a small score, tightening to <70 mg/dL from a score of 100 (or at or above the 75th percentile) and to <55 mg/dL from 1,000.
- Start lipid optimisation early — by your 30s and 40s, not your 60s. The cumulative-exposure math is the most under-appreciated piece of the field. Late intervention is highly worthwhile but late intervention loses to early intervention on absolute lifetime risk.
- Eat the Mediterranean pattern by default, and target LDL directly. See Dietary patterns. It has among the strongest hard-outcome evidence of any eating pattern: in PREDIMED, older adults at high cardiovascular risk who ate a Mediterranean diet with added olive oil or nuts had about 30% fewer heart attacks, strokes and cardiovascular deaths (HR about 0.70), a result that survived the 2013 paper's retraction and 2018 re-analysis after protocol deviations in about one in five participants, so it is moderate rather than clean-randomised evidence,[67] and a trial in people with heart disease found a benefit of similar size against a low-fat diet.[68] Its direct effect on LDL cholesterol is modest, though. To lower LDL itself, the 2026 guideline points to replacing saturated fat (red meat, butter, full-fat dairy, coconut and palm oil) with unsaturated fats and eating more fibre, nuts and legumes; a "Portfolio" diet that also added soy protein and a plant-sterol margarine lowered LDL cholesterol by about 26 mg/dL (0.67 mmol/L).[69] Don't count on supplements: in a head-to-head trial, fish oil, garlic, cinnamon, turmeric, plant sterols and red yeast rice did not lower LDL cholesterol significantly more than placebo, while a low-dose statin did.[70] Olive oil, legumes, nuts, fish, vegetables; not red and processed meat (Foods to limit).
- If you're on a statin and have muscle symptoms, work with your clinician on the n-of-1 frame from SAMSON before abandoning therapy. In people who had already quit statins, about 90% of the symptom burden was also produced by placebo. Switching statins, dose-reducing, alternate-day dosing, or running a placebo-controlled trial of one with your own doctor often salvages adherence.
- If you genuinely can't tolerate statins, bempedoic acid is now an evidence-supported alternative — 13% fewer major cardiovascular events in CLEAR Outcomes, no rise in diabetes, and muscle aches no more common than on placebo. It lowers LDL cholesterol much less than a statin and raises the risk of gout and gallstones, so mention any history of gout.
- If LDL-C remains elevated on maximally-tolerated statin (or statin plus ezetimibe), ask about a PCSK9 inhibitor. LDL-C around 30 mg/dL has so far looked safe in trials lasting up to about 8 years.[71]
- If your Lp(a) is high, treat everything else harder. No drug is approved to lower Lp(a), and the first outcomes trial of one, pelacarsen, did not significantly reduce cardiovascular events.[72] For people with high Lp(a), the National Lipid Association advises earlier and more intensive control of the other risk factors, chiefly lowering LDL cholesterol with lifestyle change and, when overall risk is high, lipid-lowering drugs, plus Lp(a) testing of parents, siblings and children.[73] If you already have cardiovascular disease and are above your LDL-C goal on the highest statin dose you tolerate, the 2026 US guideline specifically recommends adding a PCSK9 antibody.[74] Outcomes trials of the newer drugs olpasiran, lepodisiran and muvalaplin are not expected to finish until 2028–2031.
- Don't smoke; manage blood pressure and weight; maintain aerobic fitness. The lipid story sits inside the wider cardiovascular-risk frame. See Blood pressure management, VO₂ max, and Smoking and nicotine.
What's overrated
- "My LDL is normal so I'm fine." With LDL-C / apoB discordance in about one adult in five — and a higher-than-expected particle count in about one in ten — a normal LDL-C doesn't rule out high atherogenic particle burden. Get the apoB.
- The "pleiotropic effects" theory of statins. Largely settled by IMPROVE-IT, CLEAR Outcomes and the PCSK9 trials — the cardiovascular benefit tracks the fall in apoB-carrying particles, not the drug class.
- Fear of dietary cholesterol (eggs, etc.). Eating more cholesterol does raise LDL cholesterol, but only modestly, and cohort studies mostly find no clear link to cardiovascular disease at usual intakes. The American Heart Association advises focusing on the overall eating pattern rather than counting cholesterol milligrams.[75] Saturated fat is the bigger dietary driver of LDL; see Foods to limit.
- HDL cholesterol as a target. Higher HDL is no longer treated as a therapeutic goal — most HDL-raising drugs did not improve outcomes, and the one that did (anacetrapib, about 9% fewer major coronary events) produced a benefit in line with what its 18% fall in non-HDL cholesterol would predict, despite doubling HDL.[76] HDL is a marker, not a lever.
- Annual cholesterol panels as the diagnostic basis. A point-in-time LDL-C number is much less informative than the trajectory, the apoB, the Lp(a), and the calcium score together.
- Discontinuing statins because of perceived side effects without an n-of-1 trial. SAMSON showed it: in adults who had already quit statins, 90% of the symptom burden was also produced by placebo, and half were back on a statin six months after the trial. Blinded trials in 123,940 people agree that more than 90% of muscle complaints on statins are not caused by the drug.[77]
Further reading
- Ference BA, Braunwald E, Catapano AL. The LDL cumulative exposure hypothesis — evidence and practical applications. Nature Reviews Cardiology 2024.[78]
- Blumenthal RS et al. 2026 ACC/AHA/Multisociety guideline on the management of dyslipidemia. J Am Coll Cardiol 2026;87:2624.[79]
- Sayed A et al. Individual Variation in the Distribution of Apolipoprotein B Levels Across the Spectrum of LDL-C or Non-HDL-C Levels. JAMA Cardiol 2024.[80]
- Du Y et al. Impact of LDL-C and Apolipoprotein B Level Discordance and associated Lipoprotein Particle Alterations on Cardiovascular Outcomes in a large primary prevention population. Eur J Prev Cardiol 2025.[81]
- Ridker PM et al. Rosuvastatin to prevent vascular events in men and women with elevated C-reactive protein — JUPITER. NEJM 2008.[82]
- Yusuf S et al. Cholesterol lowering in intermediate-risk persons without cardiovascular disease — HOPE-3. NEJM 2016.[83]
- Nakamura H et al. Primary prevention of cardiovascular disease with pravastatin in Japan — MEGA Study. Lancet 2006.[84]
- Domanski MJ et al. Time Course of LDL Cholesterol Exposure and Cardiovascular Disease Event Risk — CARDIA. J Am Coll Cardiol 2020.[85]
- Howard JP et al. Side Effect Patterns in a Crossover Trial of Statin, Placebo, and No Treatment (SAMSON). J Am Coll Cardiol 2021.[86]
- Cannon CP et al. Ezetimibe Added to Statin Therapy after Acute Coronary Syndromes (IMPROVE-IT). N Engl J Med 2015.[87]
- Nissen SE et al. Bempedoic acid and cardiovascular outcomes in statin-intolerant patients — CLEAR Outcomes. NEJM 2023.[88]
- Sabatine MS et al. Evolocumab and clinical outcomes in patients with cardiovascular disease — FOURIER. NEJM 2017.[89]
- Kronenberg F et al. Lipoprotein(a) in atherosclerotic cardiovascular disease and aortic stenosis: a European Atherosclerosis Society consensus statement. Eur Heart J 2022.[90]
- Lp(a)HORIZON topline results — pelacarsen did not meet its primary endpoint (Novartis, September 2026).[91]
- Bohula EA et al. Evolocumab in patients without a previous myocardial infarction or stroke — VESALIUS-CV. N Engl J Med 2026.[92]