How to Prevent Atherosclerosis: A First-Principles Guide to Lipids and Plaque

I want enough data to identify particle burden, inherited risk, major accelerants, and existing disease. I do not want a giant laboratory menu that creates more noise than action. For most prevention-minded adults, the useful baseline is a standard lipid panel, non-HDL-C, ApoB, Lp(a) at least once, blood pressure, glucose or A1c, kidney function, smoking status, medication review, and a careful family history.

LDL-C and ApoB are related but not interchangeable. LDL-C estimates the amount of cholesterol carried inside LDL particles. ApoB is closer to a count of all atherogenic particles because each LDL, VLDL remnant, IDL, and Lp(a) particle carries one ApoB molecule. When ApoB and LDL-C disagree, risk often tracks more closely with ApoB, especially with insulin resistance, high triglycerides, diabetes, obesity, and treated lipid states.

Lp(a) is different again. It is a genetically determined ApoB-containing particle with additional proatherogenic, inflammatory, oxidized-phospholipid, and aortic-valve relevance. Lifestyle usually does not lower it meaningfully. The 2026 ACC/AHA guideline recommends measuring it at least once in adulthood.

Discordant results are common. The mistake is forcing them into one winner. A high ApoB and a CAC score of zero are not contradictory. One measures current particle burden; the other shows no detectable calcified coronary plaque at that moment. A high Lp(a) with a low ApoB is also not a reason to ignore Lp(a). Lp(a) carries independent inherited risk, while ApoB still helps quantify the modifiable particle burden around it.

I use disagreement to sharpen the question. Is the standard lipid panel undercounting particle risk? Is a low near-term plaque signal being mistaken for low lifetime risk? Has disease already appeared despite unremarkable current labs? Does inherited risk justify treating modifiable exposure more aggressively?

Imaging is most valuable when the result can change what you do. Coronary artery calcium, or CAC, measures calcified coronary plaque and is a strong risk reclassifier. A score of zero can lower near-term risk in many adults, but it does not prove there is no non-calcified plaque, no familial hypercholesterolemia, no Lp(a)-related risk, or no lifetime exposure problem. Younger people have also had less time to calcify plaque.

Coronary CT angiography, or CCTA, shows coronary anatomy, stenosis, and both calcified and non-calcified plaque. It has a strong diagnostic role in chest-pain pathways and can be useful selectively in prevention when anatomy will change management. It is not a universal screening test. Contrast, radiation, incidental findings, scan quality, interpretation, and downstream testing are part of the decision.

Quantitative AI plaque analysis can add measurement detail to CCTA, but it is a measurement layer, not an oracle. Vendor variability, scan quality, human confirmation, thresholds for tiny plaque volumes, and the clinical value of serial scanning remain important limitations.

Atherosclerosis is an exposure-over-time disease. Cohort data show that cumulative LDL-C during young adulthood and middle age predicts later coronary disease independently of a single midlife measurement. Human genetics tells the same story from another angle: modest lifelong reductions in LDL can produce much larger risk reductions than an identical numerical change started after decades of exposure.

Randomized trials show that lowering LDL-C reduces events. Across the Cholesterol Treatment Trialists' meta-analysis, each 1 mmol/L, or about 39 mg/dL, LDL-C reduction produced roughly a one-fifth reduction in major vascular events over the trial periods studied. The absolute benefit is larger when baseline risk is higher, but the biology does not suddenly begin at the moment a calculator crosses a threshold.

Guidelines are useful anchors, not a substitute for lifetime-risk judgment. The 2026 ACC/AHA guideline brought back LDL-C and non-HDL-C goals, uses PREVENT-ASCVD for 10- and 30-year risk, recommends earlier treatment consideration with LDL-C at least 160 mg/dL or strong premature family history in young adults, and uses CAC selectively when a treatment decision remains uncertain.

Lifestyle is foundational because it can change several causal pathways at once: ApoB, remnants, blood pressure, insulin resistance, visceral adiposity, endothelial function, smoking exposure, sleep apnea, fitness, and inflammation. The most reliable lipid lever is replacing saturated fat with unsaturated fat, not simply eating less fat. Soluble fiber, minimally processed plant foods, nuts, legumes, fish, and a Mediterranean or Portfolio-style pattern can add benefit.

Exercise often improves triglycerides, insulin sensitivity, blood pressure, fitness, and body composition more than it changes LDL-C. That does not make exercise secondary. It means exercise and lipid-lowering medication can be solving different parts of the disease model.

The ceiling matters. Lifestyle may be enough for a modest elevation in a lower-risk person. It is often not enough for LDL-C around 190 mg/dL, familial hypercholesterolemia, high Lp(a), established plaque, prior events, or a large ApoB reduction requirement. Needing medication is not a moral failure, and taking medication does not make lifestyle optional.

The medication question is not statin versus everything else. It is how much additional ApoB and LDL-C lowering is needed, which therapies have outcome evidence in a similar population, which risks matter for this person, and which regimen they can sustain.

Statins remain the foundation because they are inexpensive, potent, and supported by the broadest cardiovascular-outcomes evidence. Ezetimibe adds a complementary intestinal mechanism with low daily burden. Bempedoic acid provides oral outcome-proven therapy for statin-intolerant patients. PCSK9 antibodies produce large reductions with direct cardiovascular-outcomes evidence. Inclisiran solves a supervised-adherence problem with twice-yearly clinician administration, while its dedicated outcomes evidence is still maturing.

Lipfendra, or enlicitide, is the first oral PCSK9 inhibitor. It produced roughly 56 to 59 percent placebo-adjusted LDL-C reductions and about 50 percent ApoB reductions in Phase 3 trials. It is a potent new option, not automatic proof that statins or injections are obsolete. It must be taken in the morning on an empty stomach followed by a 30-minute wait before most food or drinks, and its product-specific cardiovascular-outcomes trial is ongoing.

Muscle symptoms deserve a real evaluation. The right response is not to dismiss the patient, and it is not to conclude that every statin at every dose is impossible after one trial. Symptoms can be related to the statin, an interaction, dose, thyroid disease, a training change, another medication, or an unrelated musculoskeletal problem. Severe pain, weakness, dark urine, fever, or systemic illness requires prompt assessment.

A structured approach can include reviewing interacting drugs and grapefruit exposure, checking the timing of symptoms, measuring CK when clinically indicated, pausing and rechallenging, trying a different statin, using a lower or intermittent dose, and combining a smaller tolerated dose with ezetimibe or another nonstatin. Pharmacogenomics can be useful in selected cases, especially with a history suggesting transporter-related myopathy, but it is not a universal prerequisite.

The goal is not loyalty to a molecule. The goal is durable reduction of causal particle exposure by a route the patient can tolerate.

After a meaningful medication or dietary change, I generally want the same lipid markers repeated after enough time to see the effect, often around 6 to 12 weeks. Once the regimen is stable and goals are met, every 6 to 12 months is common. Earlier testing may be appropriate with severe dyslipidemia, pregnancy planning, side effects, adherence questions, rapid titration, or a high-risk clinical situation.

Monitoring should answer four questions: Did the marker move by the expected amount? Is the patient actually taking the regimen as intended? Is it tolerated? Does the residual risk justify another step? For Lipfendra, administration timing is part of adherence. For injections, storage, technique, coverage interruptions, and missed doses matter. For any therapy, a smaller-than-expected response should trigger a check of adherence, secondary causes, biology, and the diagnosis before simply stacking another drug.

Routine CK or liver-enzyme testing is not necessary forever in every asymptomatic statin user. Baseline clinical context and symptom-triggered testing matter. Serial CAC is also not a treatment-response biomarker. If imaging is repeated, there should be a specific question and a realistic interval, not scan-chasing.

Many lipid decisions can be managed in primary care. Specialist input becomes especially valuable when the diagnosis, target, medication architecture, or imaging question is not straightforward. The point of referral is not to outsource prevention. It is to solve a problem that needs deeper phenotyping, combination therapy, or disease-specific judgment.

For an adult who wants a serious prevention baseline, I would begin with blood pressure, a standard lipid panel, non-HDL-C, ApoB, Lp(a) once, A1c or fasting glucose, kidney function, smoking status, medication review, and a three-generation family history focused on premature heart attack, stroke, revascularization, sudden death, and known extreme cholesterol.

If the numbers are favorable and the family history is quiet, I would focus on durable lifestyle, blood pressure, metabolic health, fitness, and periodic reassessment. If ApoB or LDL-C is high, Lp(a) is elevated, family history is strong, LDL-C is near or above 190 mg/dL, or plaque is already present, I would not wait for symptoms. I would estimate the reduction actually needed, start the lowest-burden plan likely to reach it, and verify the response.

If the treatment decision remains genuinely uncertain, CAC can be a useful tie-breaker in the appropriate age group. If symptoms or a specific anatomical question exist, CCTA or another cardiology pathway may be more relevant. I would not order every test simply because it exists.

I use a PCSK9 inhibitor in my own care. My untreated LDL-C was 192 mg/dL, my family history is severe, and statins gave me muscle symptoms. Repatha lowered my LDL-C to roughly 40 mg/dL. I do not do this because I think every adult needs the same number or the same drug. I do it because my lifetime exposure and family endpoint make the tradeoff reasonable to me.

I am excited about Lipfendra because a potent pill may remove injection friction for many people. I would not automatically switch a working, outcome-proven regimen just because the new route sounds easier. The pill has fasting choreography, product-specific outcomes are pending, and access will matter. The best treatment is the one that lowers ApoB enough, has evidence proportional to the person's risk, fits daily life, and remains tolerable and sustainable.

The central prevention mistake is waiting for atherosclerosis to become symptomatic before treating its causes. The second mistake is reducing the whole disease to one lab value. Measure the layer you are trying to change, act early enough for time to work in your favor, and close the loop with repeat data.