If you have had a cholesterol test, you were given LDL cholesterol — the “bad cholesterol” number that has anchored cardiovascular risk assessment for four decades.
There is now substantial evidence that a different measurement predicts risk more accurately, that it costs about the same, and that for a meaningful minority of people the two numbers point in opposite directions.
It is called apolipoprotein B, and understanding why it works better requires one idea about what actually causes atherosclerosis.
Particles, not cargo
Cholesterol does not float freely in blood. It travels inside lipoprotein particles, and every atherogenic particle — LDL, VLDL, IDL, lipoprotein(a) — carries exactly one molecule of apolipoprotein B.
One particle, one apoB. Measure apoB and you have counted the particles.
LDL-C measures something different: the total amount of cholesterol carried inside those particles. It is a measure of cargo, not of vehicles.
This matters because particles are not identically loaded. Some carry more cholesterol, some less. And the mechanism of atherosclerosis is thought to be particle-driven: it is the particle entering and being retained in the arterial wall that begins the process.
Two people with identical LDL-C can therefore have quite different particle counts — and it is the one with more particles who carries more risk.
What discordance analysis shows
When two measures disagree about a person, they are described as discordant. Discordance analysis compares the predictive power of two markers precisely in those individuals, which is the only way to separate highly correlated variables.
A systematic review published in 2025 compiled all such studies — 15 studies, 593,354 participants, diverse populations, with and without statin therapy.
Its conclusion was unusually direct: discordance analysis provides robust evidence that apoB is a more accurate marker of cardiovascular risk than either LDL-C or non-HDL-C, and neither LDL-C nor non-HDL-C is an adequate clinical surrogate for apoB.
A large analysis in the European Heart Journal using UK Biobank data reached the same place, with an important refinement: once apoB is accounted for, knowing LDL-C or non-HDL-C adds no further predictive information. The accompanying editorial noted that this contradicts a common belief — that non-HDL-C captures everything apoB does. It does not.
The European Atherosclerosis Society and European Society of Cardiology concluded as far back as 2019 that apoB was a more accurate marker than either alternative, and that it can be measured inexpensively, accurately and precisely.
Then why is nobody measuring it?
Adoption in clinical practice has been low, and the reasons are institutional rather than scientific.
Forty years of risk equations, trial endpoints and treatment targets are built on LDL-C. Changing the primary measurement means rebuilding all of it. Where guidelines do recommend apoB, they have often restricted it to patients with high triglycerides — on the assumption that discordance mainly arises there — which the discordance research suggests is too narrow.
In the UK, a standard NHS lipid panel gives total cholesterol, HDL, non-HDL and often LDL. ApoB is not routine, though it is available and some labs include it.
Who is most likely to be misled by LDL-C alone: people with high triglycerides, metabolic syndrome or type 2 diabetes, and people who are overweight. In these groups, particles tend to be smaller and cholesterol-depleted, so a reassuring LDL-C can sit alongside a high particle count.
There is a second reason this matters that is easy to miss: high LDL cholesterol in midlife was added to the Lancet Commission’s list of modifiable dementia risk factors in 2024, contributing 7% of population attributable risk. This is not only about heart attacks.
The honest limits
Two qualifications.
The evidence for apoB as a better predictor is strong. The evidence for apoB as a better treatment target is less established. Statins and other lipid-lowering treatments were trialled against LDL-C endpoints. That treatment lowers both is not in doubt; whether managing to an apoB target produces better outcomes than managing to an LDL-C target has not been tested head to head.
Discordance findings are nuanced. In one large 20-year cohort, elevated apoB independently predicted risk regardless of non-HDL-C — but the effect appeared mainly where LDL-C was also elevated. This is an active area, not a closed question.
What to do with this
If your lipids look normal, this is not urgent. Concordant results are the majority, and for most people LDL-C and apoB tell the same story.
Ask about apoB if you fall into the discordance-prone groups — raised triglycerides, type 2 diabetes, metabolic syndrome, central weight gain — or if you have a family history of early heart disease that your lipid panel does not seem to explain. It is a reasonable question to put to a GP, particularly as a one-off to check whether your two numbers agree.
Ask about lipoprotein(a) once, ever. Lp(a) is largely genetically determined, does not change much through life, and is not affected by diet or statins. A single measurement tells you whether you carry an inherited elevation that standard panels miss entirely. It is not routinely offered in the UK and is worth asking about if there is unexplained early heart disease in your family.
Do not order a panel of private tests on the strength of this article. More numbers without a clinician to interpret them produces anxiety, not health. One well-chosen additional measurement, discussed with your GP, is worth more than a subscription testing service.
Why this belongs on a site about ageing
Cardiovascular disease remains a leading cause of death, and the process begins decades before anything is felt. Plaque accumulates silently through your forties and fifties, and the first symptom is often the event itself.
That is what makes measurement worth taking seriously here. Unlike most things covered on this site, this one produces a number you can act on long before anything hurts.
And it illustrates something that keeps recurring: the standard measurement is not always the best one, and the gap between what the evidence supports and what is routinely done can run to a decade or more. Asking one specific question at your next appointment costs nothing.
This article is for general information and is not medical advice. Do not start, stop or change lipid-lowering medication on the basis of anything here. Discuss any testing with your GP.
Sources
- Sniderman AD, Dufresne L, Pencina KM, et al. Discordance among apoB, non–high-density lipoprotein cholesterol, and triglycerides: implications for cardiovascular prevention. European Heart Journal, 2024;45(27):2410–2418. https://academic.oup.com/eurheartj/article/45/27/2410/7663778
- Mortensen MB. ApoB triumphs once more over LDL-C and non-HDL-C in risk prediction: ready for guidelines? European Heart Journal, 2024;45(27):2419–2421. https://academic.oup.com/eurheartj/article/45/27/2419/7685937
- ApoB, LDL-C, and non-HDL-C as markers of cardiovascular risk: a systematic review of discordance analyses. Journal of Clinical Lipidology, 2025. https://www.lipidjournal.com/article/S1933-2874(25)00315-0/abstract
- Concordance–discordance between apolipoprotein B and lipid biomarkers in predicting 20-year atherosclerotic cardiovascular disease risk: the ATTICA study. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12434434/
- Livingston G, et al. Dementia prevention, intervention, and care: 2024 report of the Lancet standing Commission. The Lancet, 2024.
This article is for general information and is not medical advice. If a health problem is affecting your daily life, speak to your GP.
Comments
We read every comment. Be kind, stay on topic, and please don't ask for medical advice — we can't give it.