Most coronary plaque begins when apoB-containing particles enter the artery wall and stay there long enough to cause trouble.

Patients are still taught a cartoon version of lipid biology, where LDL plays the villain, HDL the hero, and statins are framed as either salvation or overreach. It is tidy, memorable, and wrong in the ways that matter. The patients who expose the weakness in that story are the ones who look respectable on a standard panel: the person with an LDL-C that is not especially dramatic but an apoB that is high, the person with premature coronary disease running through the family and an elevated Lp(a), or the person with a handsome HDL-C who still ends up with plaque.

The cleaner biological story is that coronary artery disease is driven by cumulative exposure to apoB-containing particles. LDL is the dominant apoB-containing particle in most people, but remnants and lipoprotein(a), or Lp(a), also contribute. Cholesterol matters because it is cargo carried inside those particles, and the artery is exposed to the particles carrying it rather than to an abstract cholesterol number (Sniderman 2019; Ference 2017; Borén 2020).

That shift in framing solves two problems at once. It explains why apoB can be more informative than LDL-C, particularly in insulin resistance, hypertriglyceridaemia, and mixed dyslipidaemia where the cholesterol content of each particle varies (Sniderman 2019; De Oliveira-Gomes 2024). It also explains why statins matter so much without forcing us to pretend they are the only drugs capable of reducing coronary events.

That matters even more in 2026, because the new AHA/ACC multisociety dyslipidaemia guideline has nudged the clinical language closer to the underlying biology. LDL-C still carries most of the treatment conversation, but the official update tightens very-high-risk secondary-prevention goals to LDL-C below 1.4 mmol/L [55 mg/dL], brings back non-HDL-C goals, and recommends a once-in-lifetime Lp(a) measurement in adults (Blumenthal RS 2026; American College of Cardiology 2026; American Heart Association 2026).

‍

The case: plaque starts with particle retention

Atherosclerosis begins when apoB-containing particles cross the endothelium, are retained in the subendothelial space, become modified, and start a chronic inflammatory repair process that eventually looks like plaque (Sniderman 2019; Borén 2020). Ference and colleagues pulled together cohort data, Mendelian randomisation, and randomised trials covering more than 2 million participants and more than 150,000 cardiovascular events, and the relationship between LDL exposure and atherosclerotic disease was dose-dependent, log-linear, and stronger with longer duration of exposure (Ference 2017). In plain English, the artery keeps score over time.

Borén’s 2020 consensus statement gives the mechanism more texture. LDL particles cross the endothelium, bind to arterial-wall proteoglycans, and become trapped. Once retained, they are chemically modified and inflammatory signalling is amplified (Borén 2020). Moore’s macrophage review and Khatana’s oxidized LDL review complete the sequence: monocytes are recruited into the intima, macrophage foam cells form, and advanced lesions develop necrotic cores, fibrous-cap thinning, and thrombosis (Moore 2011; Khatana 2020). The disease is what those particles are doing inside the artery wall.

‍

Why apoB is often closer to the biology than LDL-C

Most clinically relevant atherogenic particles in adult lipid assessment carry a single apoB molecule, making apoB a practical estimate of particle number. LDL-C tells us how much cholesterol mass those particles are carrying. When the particles are relatively cholesterol-poor, a patient can have more atherogenic traffic than the LDL-C number suggests (Sniderman 2019). That is why apoB is often closer to the biology when LDL-C and particle number are discordant, and one reason a patient with apparently reasonable LDL-C can still carry residual risk.

The large epidemiology is consistent with that model. In the Emerging Risk Factors Collaboration analysis of 68 prospective studies, including 302,430 people without baseline vascular disease, the hazard ratio for coronary heart disease was 1.49 [95% CI 1.39 to 1.60] for the apoB/apoA-I ratio, compared with 1.38 [1.09 to 1.73] for directly measured LDL-C (Emerging Risk Factors Collaboration 2009). In the same dataset, HDL-C was inversely associated with coronary risk, hazard ratio 0.78 [0.74 to 0.82], which is useful epidemiologically but does not settle the causal question.

The modern discordance data are what make apoB especially useful in clinic. De Oliveira-Gomes and colleagues summarised UK Biobank analyses showing that about 18 percent of statin-naive adults without cardiovascular disease had materially discordant apoB and LDL-C values. In that discordant group, apoB still predicted atherosclerotic cardiovascular risk, with an adjusted hazard ratio per standard deviation of 1.23 [95% CI 1.12 to 1.35], while LDL-C did not (De Oliveira-Gomes 2024). If LDL-C looks respectable but apoB remains high, I would assume LDL-C is understating particle burden.

ApoB is also where the treatment story becomes more coherent. In a meta-analysis of 29 randomised trials including 332,912 participants, Khan and colleagues found that each 0.10 g/L [10 mg/dL] reduction in apoB was associated with lower all-cause mortality, relative risk 0.95 [95% CI 0.92 to 0.99], lower cardiovascular mortality, 0.93 [0.88 to 0.98], and lower major adverse cardiovascular events, 0.93 [0.90 to 0.97] (Khan 2020). That does not prove every lipid drug is interchangeable. It does support the broader point that lowering apoB-containing particle burden is a meaningful mechanism across therapies.

Guidelines have not fully surrendered to particle language, but the 2026 update is more serious about residual atherogenic burden. In very-high-risk ASCVD, the official ACC synopsis gives treatment goals of LDL-C below 1.4 mmol/L [55 mg/dL], non-HDL-C below 2.2 mmol/L [85 mg/dL], and apoB below 0.65 g/L [65 mg/dL] when it is measured (Blumenthal RS 2026; American College of Cardiology 2026). That does not mean every clinic will suddenly speak apoB first. It does mean the guideline is less willing to settle for a merely respectable LDL-C when residual particle burden is likely to remain.

When apoB is unavailable, non-HDL-C is the most practical bridge marker because it captures cholesterol across all apoB-containing particles, including VLDL and remnants, rather than LDL alone. That matters most in mixed dyslipidaemia and insulin resistance, where remnant burden can be clinically important and LDL-C can look calmer than the particle story really is. The return of explicit non-HDL-C goals in the 2026 guidance is sensible for that reason, even if apoB remains the cleaner measure of atherogenic burden (American College of Cardiology 2026; American Heart Association 2026).

‍

The metabolic layer: insulin resistance, visceral fat, and the TG:HDL ratio

Coronary disease is still an apoB story, but insulin resistance and central adiposity change the metabolic setting on which that apoB burden acts. The usual pattern is increased hepatic VLDL production, higher triglycerides, lower HDL-C, more remnant-rich lipoproteins, and LDL particles carrying less cholesterol each. That is one reason LDL-C can look less alarming than apoB in the very people whose metabolic picture is most atherogenic (Ormazabal 2018; De Oliveira-Gomes 2024).

Insulin resistance is more informative than fasting insulin alone. In adults without diabetes, high versus low fasting insulin carried a CHD risk ratio of 1.12 [95% CI 0.92 to 1.37], while HOMA-IR carried a clearer association at 1.64 [1.35 to 2.00] (Gast 2012). Mechanistically, insulin resistance increases free-fatty-acid flux to the liver and promotes VLDL overproduction, which then feeds triglyceride-rich remnants and a more discordant apoB-LDL picture (Ormazabal 2018). A Mendelian randomisation analysis using an insulin-resistance phenotype instrument found CAD odds of 1.79 [1.57 to 2.04] and MI odds of 1.78 [1.54 to 2.06] per standard deviation higher genetically predicted insulin resistance (Chen 2020). If I mention hyperinsulinaemia, I would treat it as a clue to insulin resistance rather than a separate treatment target.

The TG:HDL ratio sits in the same bucket because it is often the visible footprint of that VLDL-remnant state. In overweight or obese adults with type 2 diabetes, the high-TG/low-HDL phenotype predicted CAD with a hazard ratio of 1.48 [95% CI 1.14 to 1.93] (Kaze 2021). In broader CHD cohorts, a higher TG:HDL ratio tracked all-cause mortality [HR 2.92, 95% CI 1.75 to 4.86] and major adverse cardiovascular events [1.56, 1.11 to 2.18] (Guan 2022). In UK Biobank, the highest versus lowest TG:HDL quartile predicted incident CVD with a hazard ratio of 1.29 [1.23 to 1.36], but 40 percent of that signal was mediated by dyslipidaemia, with further mediation by diabetes and hypertension (Che 2023). That makes the ratio a useful cheap flag for metabolic dysfunction and likely remnant excess, not a replacement for apoB.

Visceral fat is probably the most useful body-composition clue because it sits closer to the biology than BMI does. More visceral and ectopic fat means a metabolically noisier adipose organ, more insulin resistance, more hepatic substrate delivery, and usually more VLDL traffic (Gruzdeva 2018; Ormazabal 2018). In prospective data, CT-measured visceral adipose tissue predicted coronary calcium progression [HR 1.004, 95% CI 1.001 to 1.007] and mediated 51.8 percent [44.5 to 58.8] of the insulin-resistance/adipose-dysfunction pathway, although the outcome was calcium progression rather than myocardial infarction or cardiovascular death (Antonio-Villa 2023). I would not order CT just to hunt for coronary risk, but a large waist, central obesity, and a rising TG:HDL ratio together are often telling you to look for remnant-rich dyslipidaemia and apoB-LDL discordance.

The practical response is not to turn visceral fat or insulin resistance into a new dogma. It is to treat them as prompts to look harder at the whole cardiometabolic phenotype, especially apoB, blood pressure, glucose, HbA1c, and weight trajectory.

‍

Lp(a): inherited residual risk that deserves more respect

Lp(a) sits inside the apoB story, but it is different enough to deserve its own section. It is an apoB-containing particle with an attached apolipoprotein(a) component and a heavy oxidised phospholipid burden. Its level is largely genetically determined, varies substantially by ancestry, and is only modestly altered by lifestyle. That is why a patient can be doing many sensible things and still carry considerable inherited residual risk (Reyes-Soffer 2022; Kronenberg 2022).

The American Heart Association scientific statement by Reyes-Soffer and colleagues is a good anchor because it is sober and clinically useful. The statement summarised population data showing an approximately log-linear increase in atherosclerotic cardiovascular risk above the population median, with a hazard ratio of 1.11 [95% CI 1.10 to 1.12] per 50 nmol/L higher Lp(a), independent of traditional risk factors (Reyes-Soffer 2022). In plain English, Lp(a) is a common inherited contributor to risk that routine lipid discussions can easily understate.

Björnson’s 2024 genetic analysis sharpens the point. In 377,572 UK Biobank participants, the hazard ratio for coronary heart disease per 50 nmol/L apoB was 1.47 [95% CI 1.36 to 1.58] for the Lp(a) cluster, compared with 1.04 [1.02 to 1.05] for the LDL cluster. The estimated per-particle atherogenicity of Lp(a) was about 6.6-fold [95% CI 5.1 to 8.8] greater than LDL (Björnson 2024). That goes a long way toward explaining why some people declare trouble early despite an LDL-C that never looked theatrical.

Kronenberg’s review is helpful for practice because it captures where guideline thinking was already moving. The 2026 multisociety dyslipidaemia guideline now makes that explicit, recommending that adults have Lp(a) measured once in a lifetime (Blumenthal RS 2026; American College of Cardiology 2026; American Heart Association 2026). The practical implication is not that Lp(a) replaces LDL-C or apoB as the main modifiable target. It means that when Lp(a) is high, the rest of risk-factor control usually deserves a more serious tone.

‍

HDL-C: useful marker, poor therapeutic north star

This is the part that still bothers people, because many were taught for years that a high HDL-C was a kind of cardiovascular charm. Observationally, low HDL-C is certainly associated with higher risk. In the Emerging Risk Factors Collaboration dataset, HDL-C was inversely associated with coronary disease, hazard ratio 0.78 [95% CI 0.74 to 0.82] (Emerging Risk Factors Collaboration 2009). But observational association is not the same thing as a good treatment target.

The drug-trial history is the correction. In Keene’s meta-analysis of 39 randomised trials including 117,411 participants, niacin, fibrates, and CETP inhibitors all raised HDL-C, yet none produced a significant all-cause mortality benefit: niacin odds ratio 1.03 [95% CI 0.92 to 1.15], fibrates 0.98 [0.89 to 1.08], CETP inhibitors 1.16 [0.93 to 1.44] (Keene 2014). Niacin reduced non-fatal myocardial infarction only in pre-statin era trials, odds ratio 0.69 [0.56 to 0.85], and that effect largely disappeared once background statin therapy became standard, odds ratio 0.96 [0.85 to 1.09] (Keene 2014).

The broader review literature reaches the same conclusion from different angles. Von Eckardstein’s state-of-the-art review argues that HDL function is biologically interesting but HDL-C concentration is a crude surrogate, and larger Mendelian-randomisation studies have failed to show a significant association between genetically higher HDL-C and ASCVD risk (von Eckardstein 2023). Franczyk’s review adds the counterintuitive point that very high HDL-C is not automatically reassuring. In pooled Japanese cohorts, HDL-C of at least 2.3 mmol/L [90 mg/dL] was associated with higher atherosclerotic cardiovascular mortality, hazard ratio 2.37 [95% CI 1.37 to 4.09] (Franczyk 2021). The torcetrapib experience is a useful caution here: large HDL-C increases did not translate into better outcomes, and cardiovascular events, hazard ratio 1.25 [1.09 to 1.44], and all-cause mortality, 1.58 [1.14 to 2.19], were higher in trial data reviewed by Franczyk (Franczyk 2021). That is exactly the point: pharmacologically raising HDL-C has not reliably improved outcomes, and HDL-C concentration is not the same thing as HDL function.

So I would treat HDL-C as context, not as a hero. A low HDL-C can still flag insulin resistance, hypertriglyceridaemia, central adiposity, or metabolic dysfunction. A high HDL-C should not make a doctor or a patient complacent.

‍

Are statins essential?

This depends on what work the word essential is being asked to do. If the claim is that statins are the only class with meaningful coronary outcome data, that is plainly wrong. If the claim is that statins remain the best-supported and most practical first-line therapy for many patients, that is still true.

The statin evidence is deep for a reason. In the Cholesterol Treatment Trialists’ meta-analysis of 26 randomised trials including 169,138 participants, each 1.0 mmol/L reduction in LDL-C was associated with a relative risk of 0.78 [95% CI 0.76 to 0.80] for major vascular events and 0.90 [0.87 to 0.93] for all-cause mortality (CTT Collaboration 2010). In the Cochrane primary prevention review of 18 trials and 56,934 participants, statins reduced all-cause mortality, odds ratio 0.86 [95% CI 0.79 to 0.94], combined fatal and non-fatal cardiovascular disease, relative risk 0.75 [0.70 to 0.81], stroke, 0.78 [0.68 to 0.89], and revascularisation, 0.62 [0.54 to 0.72] (Taylor 2013). That is why guidelines still put statins first (Grundy 2019).

Where the discussion becomes more interesting is what happens after that. In the 2022 BMJ network meta-analysis of 83,660 statin-treated adults, ezetimibe added to statin therapy reduced myocardial infarction, relative risk 0.87 [95% CI 0.80 to 0.94], and stroke, 0.82 [0.71 to 0.96]. PCSK9 inhibitors reduced myocardial infarction, 0.81 [0.76 to 0.87], and stroke, 0.74 [0.64 to 0.85], without a clear short-term mortality benefit over available follow-up (Khan 2022). Those are clinically meaningful outcome signals.

The older-patient data also argue against treating statins as the only route to benefit. Gencer and colleagues analysed 244,090 participants, including 21,492 adults aged 75 years or older, and found that LDL-C lowering reduced major vascular events by 26 percent per 1 mmol/L reduction, relative risk 0.74 [95% CI 0.61 to 0.89]. The benefit did not differ significantly between statin and non-statin therapy, interaction P=0.64 (Gencer 2020).

Bempedoic acid matters because it addresses the practical counterargument of statin intolerance. In 779 high-risk patients on maximally tolerated statins, Goldberg’s CLEAR Wisdom trial showed an LDL-C difference of -17.4 percent [95% CI -21.0 to -13.9] and an apoB difference of -13.0 percent [95% CI -16.1 to -9.9] at 12 weeks (Goldberg 2019). In the JAMA primary-prevention subgroup analysis by Nissen and colleagues, involving 4,206 statin-intolerant high-risk patients without prior cardiovascular events, the primary endpoint occurred in 5.3 percent with bempedoic acid versus 7.6 percent with placebo, hazard ratio 0.70 [95% CI 0.55 to 0.89], and all-cause mortality fell with a hazard ratio of 0.73 [0.54 to 0.98] (Nissen 2023).

Ezetimibe combination therapy is also harder to dismiss now. In the RACING post hoc analysis of very-high-risk ASCVD, involving 1,511 of the trial’s 3,780 patients, moderate-intensity statin plus ezetimibe produced a primary endpoint hazard ratio of 0.96 [95% CI 0.71 to 1.30], lower LDL-C at one year [1.5 vs 1.7 mmol/L; 57 vs 65 mg/dL], and fewer intolerance-related dose reductions or discontinuations [4.6 percent vs 7.7 percent] (Lee 2023). The 2025 individual patient data meta-analysis by Lee and colleagues, covering 8,180 patients with ASCVD, found a similar pattern, hazard ratio 0.98 [95% CI 0.84 to 1.15], with less new-onset diabetes and less intolerance-related dose reduction (Lee 2025).

So the serious position here is straightforward. Statins remain the preferred base because the randomised outcomes evidence is deepest, they are inexpensive, and they work. But the therapeutic principle is lowering cumulative exposure to apoB-containing particles. Ezetimibe, PCSK9 inhibition, and bempedoic acid all matter because they help achieve that goal in different clinical situations (Khan 2020; Khan 2022; Grundy 2019).

‍

What cuts the other way

A few cautions matter if we want to stay serious. Most non-statin hard-outcome evidence is add-on evidence rather than clean statin-replacement evidence, so it would be sloppy to present these therapies as universally interchangeable. ApoB is mechanistically cleaner, especially when LDL-C and particle number are discordant, but most guidelines and many older outcome trials are still written in LDL-C language, which is why practice has not completely shifted over to particle-based thinking (Grundy 2019; De Oliveira-Gomes 2024). HDL biology remains more interesting than HDL-C concentration, so saying HDL does not matter would be too blunt (von Eckardstein 2023). And several of the most useful causal framing papers in this field are scientific statements, consensus statements, or narrative reviews rather than primary randomised outcome trials (Ference 2017; Borén 2020; Reyes-Soffer 2022).

The metabolic markers above sit in that same caution zone. Fasting insulin, HOMA-IR, TG:HDL, and visceral-fat measures are useful risk markers and phenotype clues, but they are not magic treatment targets. The strongest hard-outcome evidence in this article still comes from apoB-lowering therapies.

‍

What this means in clinic

The practical sequence is less dramatic than the internet makes it sound. I still care about LDL-C, because most treatment pathways still revolve around it. I also want to know when LDL-C is understating atherogenic traffic, which is where apoB helps, particularly in insulin resistance, type 2 diabetes, hypertriglyceridaemia, mixed dyslipidaemia, premature coronary disease, or apparently discordant results (Sniderman 2019; De Oliveira-Gomes 2024). A high TG:HDL ratio or a large waist is another clue that the metabolic picture deserves a closer look. If apoB is unavailable, non-HDL-C is the next best bridge, especially when triglycerides are up and remnants are likely to matter.

I also think adults should have Lp(a) measured once in adult life (I measure more often than this in patients taking PCSK9 inhibitors). That is now guideline-backed rather than merely fashionable, and it matters because inherited residual risk is common enough to surprise us and important enough to change how aggressively we manage everything else that remains modifiable (Blumenthal RS 2026; American College of Cardiology 2026; American Heart Association 2026).

If drug therapy is warranted, statins remain the usual first step because they reduce events, are widely available, and still have the cleanest evidence base. If risk remains high, the newer target language matters. In very-high-risk ASCVD, the 2026 guideline aims for LDL-C below 1.4 mmol/L [55 mg/dL], with non-HDL-C below 2.2 mmol/L [85 mg/dL] as a practical companion target and apoB below 0.65 g/L [65 mg/dL] when it is measured (Blumenthal RS 2026; American College of Cardiology 2026). If true statin intolerance limits what you can do, bempedoic acid now offers more than a theoretical alternative. What I would not do is chase HDL-C upward as though the number itself were a therapeutic destination.

This is also where shared decision-making matters. The right choice depends on absolute risk, whether we are talking about primary or secondary prevention, the patient’s preferences, prior side effects, and the practical realities of cost and access. In borderline or intermediate primary prevention, coronary artery calcium can help break ties: a CAC of 0 can justify holding off in selected patients, while CAC 1-99 favours treatment in older adults and CAC of at least 100 makes treatment harder to avoid (American College of Cardiology 2026). A person with established ASCVD and recurrent events deserves a different threshold for treatment intensity than a lower-risk primary-prevention patient who is wary of medication and has had troublesome adverse effects before.

Patients generally understand this very well when it is explained plainly. The clinical task is to reduce the cumulative burden of atherogenic particles over time, using the safest and most practical combination of tools the patient can tolerate.

‍

What we still do not know

The biggest unanswered question in this space is not whether lowering LDL or apoB helps. That case is already strong. Within the source set used here, the bigger unresolved question is whether large dedicated reductions in Lp(a) will translate into event reduction independent of standard apoB lowering. Until those outcomes data are clearly established, the most defensible position is that elevated Lp(a) identifies inherited risk that should push us to manage every other modifiable factor properly rather than admire a reassuring HDL-C and hope for the best.

‍