The ApoB Test, and Whether to Ask for One

Your cholesterol panel came back fine and someone has told you to ask for an ApoB test. LDL cholesterol measures how much cholesterol is being carried. ApoB counts the particles carrying it. They usually agree, and when they do not it can run either way. The direction this article is about is a high ApoB behind a normal-looking LDL. In 2,794 American adults followed 25 years, that was 8.6 percent [1]. Every other cohort drew its line differently and got a different number.

What this comes down to

Each line is unpacked below, with its source.

1. LDL cholesterol measures the cholesterol being carried. ApoB counts the particles carrying it, because every one of those particles has exactly one ApoB protein on it.
2. Nobody can tell you how often the two disagree, because the fraction is made by where the line is drawn. One Polish survey got anything from 0.2 to 45.2 percent out of the same people.
3. The group that matters, a high ApoB behind a normal-looking LDL, was 8.6 percent in one American cohort. It was 5.3 percent in a Korean one, and 5 percent as a Greek table prints it. Three different definitions, so they are not pooled here.
4. In that group risk ran higher in four of the cohorts here. A fifth, MESA, found it for calcium progression but not consistently for other endpoints, and a sixth did not find it at all. Every one of those studies is observational.
5. No guideline body has a trial-backed ApoB target. Europe does print ApoB goals, under 65, 80 and 100 mg/dL by risk group, but calls them secondary to LDL cholesterol. America lists it as a risk flag with no grade. Canada declines to set one and says why. The UK does not mention it.

What the ApoB test measures, and what LDL cholesterol does not

Think of your bloodstream as a delivery network. LDL cholesterol tells you how much cargo is moving. ApoB tells you how many trucks are on the road.

That is not a metaphor stretched for effect. Every atherogenic particle carries exactly one apolipoprotein B molecule, apoB-100 on LDL, IDL, VLDL and Lp(a), apoB-48 on chylomicron remnants. So counting ApoB counts particles, which is why the European guideline describes the disease as caused by LDL and other ApoB-containing lipoproteins [11]. A standard lipid panel does not count anything. It estimates the cholesterol content of one class of them.

Most of the time, more cargo means more trucks and the two numbers move together. The interesting case is the small truck. Someone can carry a lot of particles that are each lightly loaded, and their LDL cholesterol will look unremarkable while their particle count is high.

Whether that distinction earns a place on your panel is the rest of this article. The answer is more interesting than either side of the argument usually admits.

How often do ApoB and LDL cholesterol actually disagree?

You will read that they disagree in about one in five people. I am not going to tell you that, because I went looking for where it comes from and it is not supported.

The number has no single primary source that I could reach. The nearest thing to one is a 2003 paper by Sniderman and colleagues, passed along second hand ever since, cited alongside the American cohort above. The Korean cohort's own introduction repeats it as 10 to 30 percent of individuals, without measuring it [2]. I could not trace it further through the European guideline's reference list, which the publisher would not serve me. I did read that guideline's own recommendation and target tables.

What the primary cohorts show is both smaller and more useful. In 2,794 American adults followed for 25 years, 8.6 percent had a high ApoB with LDL cholesterol below the cohort median [1]. In 14,205 Korean men at a corporate health checkup, reported in 2021, it was 5.3 percent [2]. That paper carries no funding statement at all, and it's the only source here for that number. In the Greek ATTICA cohort, measured against a clinical threshold rather than a median, the table prints 5 percent: 100 people of the 2,208 that table classifies [3].

Those three numbers are not comparable to each other, and I have deliberately not averaged them. Each cohort drew its dividing line somewhere different, and that choice is most of what the number is.

Why does every source give a different percentage?

Because discordance is not a property of your blood. It is a property of the line somebody drew through a population.

The clearest demonstration is a Polish national survey that ran the same people through several different sets of dividing values. The share called discordant ranged from 0.2 percent to 45.2 percent [5]. Same blood, same arithmetic, different cut points.

The Greek cohort shows it twice over, in two tables about the same people. Measured against clinical thresholds, 537 of the 2,208 adults that table classifies fell in a discordant cell, which is about 24 percent. Ranked by fifths instead, the paper reports 48.7 percent concordant, which is 51.3 percent discordant [3]. Both figures come out of the same cohort on the same day, by two different rules.

So when a clinic tells you that one in five people are discordant, the honest reply is: discordant by whose definition? The Polish survey's authors put it plainly in their own abstract: results from different discordance approaches cannot be directly compared [5].

What happened to people whose ApoB was high and LDL looked fine?

This is the question worth asking, and the evidence here is real but uneven.

Take 27,533 American women, followed a median of 17 years [4]. That group carried an age-adjusted hazard ratio of 2.48 for coronary events, against concordant women below the median. After full adjustment the gap narrows to about 20 to 50 percent, in the published Circulation abstract's own words. The first author declares research grant support from AstraZeneca and a lipid diagnostics company, and consulting for Pfizer, Quest Diagnostics and Genzyme.

In the 25-year American cohort, that group had an odds ratio of 1.55 for any coronary calcium on a scan [1]. Calcium on a scan is a picture of plaque, not a heart attack. In the Korean cohort, calcium progression carried a relative risk of 1.26 [2]. In 375,544 UK adults on no lipid-lowering therapy, major cardiovascular events carried a hazard ratio of 1.11 [6]. That paper's own funding statement could not be read.

And then the result that does not fit. The Greek cohort used the threshold a laboratory would actually print on your report, and followed 1,988 people 20 years to real events. Its hazard ratio was 1.320, with an interval from 0.665 to 2.620 [3]. That interval cannot rule out either a real effect or none at all. The group behind it is the 100 people its threshold table counts, minus whoever was lost along the way, and the paper does not print that number.

A 2020 MESA analysis found ApoB discordance inconsistently associated with coronary calcium. Though in that same analysis, a discordantly high ApoB was associated with calcium progression [7]. So of the cohorts here, four found higher risk outright. MESA found it for one endpoint and not others, and the Greek cohort did not find it at all. Anyone citing only the ones that found it is choosing their evidence, and so is anyone counting MESA cleanly either way.

The mirror image is quieter and more consistent. Where LDL cholesterol is high and ApoB is low, no cohort found excess risk. The comparisons behind that are not interchangeable. In the largest of them the reference group is concordant women above the median, while the others compare against concordant low. The same finding appears in opposite arithmetic.

Is ApoB actually the better number, or just the newer one?

The strongest argument for it does not come from the discordance literature at all. It comes from genetics.

A 2020 study used inherited genetic variants as a stand-in for lifelong exposure [8]. It took genetic markers of all three measures, ApoB, LDL cholesterol and triglycerides, from roughly 440,000 UK Biobank adults. Those were then tested against coronary disease in a separate 60,801 cases and 123,504 controls. With all three in one model, ApoB carried an odds ratio of 1.92 for coronary disease. LDL cholesterol came out at 0.85, with an interval from 0.57 to 1.27. On its own, LDL cholesterol looked just as predictive.

That is suggestive and it is not proof. An interval that wide means the model could not separate the two, and the authors say so themselves. It is also about lifelong genetically proxied exposure, which is not the same as what happens when you change something this year.

A 2011 meta-analysis of 12 reports covering 233,455 people put ApoB ahead of both [9]. The standardized relative risk ratios were 1.43 for ApoB, 1.34 for non-HDL cholesterol and 1.25 for LDL cholesterol. Its lead author, Allan Sniderman, has argued the case for ApoB in print for two decades. He is also a co-author on two of the other studies cited here. That meta-analysis's own funding statement is not readable by any route I tried.

Take a 2022 analysis of 389,529 adults on no lipid-lowering drugs [10]. With all three measures assessed together, ApoB carried a hazard ratio of 1.27 for a first heart attack. Two of its three cohorts came from trials funded by drug companies, and its authors carry industry grants. Those are standardized units the paper does not convert, so neither will I.

One name is worth knowing. Brian Ference reports grants and advisory fees from Amgen, Merck, Novartis, Regeneron, Sanofi, Pfizer, Esperion and about a dozen more. He is both a co-author on that genetic paper and a named task-force member on the European guideline this article quotes. That is not an accusation. It is the shape of a small field, and you should be able to see it.

What do the guidelines actually say about ApoB?

Less than the internet suggests, and they disagree with each other.

Europe recommends measuring it. The 2019 guideline recommends ApoB analysis for risk assessment, particularly in people with high triglycerides, diabetes, obesity, metabolic syndrome or very low LDL cholesterol. It adds that ApoB can be used as an alternative to LDL cholesterol, if available [11]. That carries the evidence grade the same document defines as consensus of expert opinion, small studies, retrospective studies or registries. The 2025 focused update did not change it.

America names it without grading it. ApoB appears in a table of risk-enhancing factors, under a heading that begins "If measured" [12]. A level at or above 130 mg/dL is described there as a risk-enhancing factor. It carries no class of recommendation and no level of evidence anywhere in the document.

Canada goes furthest. Its 2021 guideline recommends that above a triglyceride of 1.5 mmol/L, non-HDL cholesterol or ApoB be used instead of LDL cholesterol for screening. It grades that a strong recommendation on high-quality evidence [13].

The United Kingdom does not mention ApoB at all. Its current guideline names apolipoprotein zero times and uses non-HDL cholesterol instead [14]. That is an absence rather than a rejection, and I could not find the committee's reasoning.

The thing all four share is more important than where they differ. Not one of them has a trial-backed ApoB target. Europe prints secondary ApoB goals of under 65, 80 and 100 mg/dL for very-high, high and moderate-risk people [11]. The word to notice is secondary: the primary goal there is still LDL cholesterol. Canada declines to set one, and says why [13]. To date, no clear target to which LDL cholesterol, non-HDL cholesterol or ApoB levels should be lowered is clearly identified in randomized trials. America sets none. So a number presented to you as the one to aim for is, at best, a secondary European goal for a risk group you may not be in.

Should you ask for an ApoB test?

Here is how I would decide, and I will say plainly which parts are evidence and which are my practice.

The European guideline recommends measuring it if your triglycerides are high, or you have diabetes, obesity or metabolic syndrome, or your LDL cholesterol is very low [11]. That rests on expert consensus rather than trial evidence, and it's where the two numbers are most likely to part company.

The evidence does not support treating it as a target to drive down. Europe's ApoB numbers are secondary goals, and Europe describes them as defined by inference rather than extensively studied in trials. Canada declines to set one at all. It does not support dropping the rest of your lipid panel. Canada does prefer ApoB or non-HDL cholesterol over LDL for screening, but only when triglycerides are above 1.5 mmol/L. And it cannot tell you what happens to your risk if you change your own number this year, because none of this research followed anyone doing that.

My own practice, which is not the same as evidence. I want it on the panel once, early, for anyone whose numbers look reassuring while the rest of their picture does not. Not because the particle count is magic. It is one more assay on blood you're already giving, and a normal-looking panel is the single most common reason a person stops asking questions. If you want the longer version of that argument, I wrote it in why your bloodwork can come back normal while you feel terrible.

One thing to know before you ask. Your results come back in different units depending on where you are. Milligrams per deciliter in some places, grams per liter in others, and the conversion is not something the sources print. Ask your physician to read your number against the reference your laboratory used, not against one you found online.

It is completely bio-individual, and it is not a cookie cutter approach. Two people with the same ApoB will have two different reasons for it.

What I would do with a panel where the two disagree

Not panic, and not reorder it three times.

I'll tell you my own, from a coaching call in May 2026, because it is the only lipid story I can tell you first hand. I am a hyper-responder to cholesterol. It runs in my family. So I ran an experiment on myself: a month with no red meat, bloodwork before and after, and I also increased my fiber. My cholesterol came down and my ApoB came down with it, and I was genuinely surprised by how much. I am not printing a percentage. I have told that story several times and the number has moved each time, and a figure I cannot stand behind is worse than no figure.

What I take from it is the shape, not the size. A single month of one deliberate change moved both numbers in the same direction in a person who responds strongly. That is one person, with no control group and no blinding. The results presented reflect one individual's outcome and should not be interpreted as typical.

So the practical sequence. Get the number once, with the rest of the panel, on the same draw. Read it against your laboratory's own reference, not an internet threshold. If it disagrees with your LDL cholesterol, that is information about which number is doing the work for you, not a diagnosis. And take it to your physician, because every finding in this article is an association in a population, and none of it was measured on you.

For the nearest thing to a controlled experiment on yourself, what heart rate variability actually measures covers how to do it on a different number. Gamifying your blood glucose is the same discipline applied to a third. The habit transfers even when the marker does not.

Actionable steps

If you take one thing from this article, take this list.

1. Ask for it on the same draw as your usual panel. It is one more assay, not another appointment.
2. The clearest case for it: high triglycerides, diabetes, obesity, metabolic syndrome, or a very low LDL cholesterol.
3. Read it against the reference range your own laboratory printed, in your own laboratory's units.
4. Treat a disagreement between ApoB and LDL cholesterol as information about which number to watch, never as a diagnosis.
5. Do not chase a target of your own choosing. The only ApoB numbers printed as goals are Europe's secondary ones, and Europe says itself they were set by inference rather than extensively studied in trials.
6. This article is education, not medical advice. Take your own numbers to your own physician.

Where to take this next

If your panel came back normal and you want to know which of its numbers is actually answering your question, that is what the assessment is for. Discover your science, then optimize your life.

Quick answers

What is the difference between ApoB and LDL cholesterol?

LDL cholesterol measures how much cholesterol is being carried inside one class of particle. ApoB counts the particles themselves, because every atherogenic particle carries exactly one apolipoprotein B molecule. Most of the time the two move together. When they part company, it is usually because someone is carrying many lightly loaded particles, so the cargo measurement looks ordinary while the particle count is high.

How often do ApoB and LDL cholesterol disagree?

There is no single answer, and anyone who gives you one has not looked. The fraction depends entirely on where the dividing lines are drawn. One Polish national survey produced anything from 0.2 percent to 45.2 percent out of the same people depending on the cut points used [5]. For the group that matters, a high ApoB with a normal-looking LDL cholesterol, individual cohorts reported 8.6 percent [1], 5.3 percent [2] and 5 percent [3]. Those use three different definitions and should not be averaged.

Is a high ApoB with normal LDL cholesterol dangerous?

It is associated with higher risk in most but not all of the cohorts that looked. Four of the cohorts here found higher risk outright. That includes a hazard ratio of 2.48 for coronary events in 27,533 women, before full adjustment [4]. That 2014 paper's first author declares drug-company grants and consultancies. The same group carried a hazard ratio of 1.11 for major cardiovascular events in 375,544 UK adults in 2025 [6]. That paper's own funding statement could not be read. Two did not find it outright. A Greek cohort followed 20 years to real events gave 1.320, with an interval from 0.665 to 2.620, in 100 people [3]. A 2020 MESA analysis found ApoB discordance inconsistently associated with coronary calcium, though in that same analysis a discordantly high ApoB was associated with calcium progression [7]. All of it is observational, and none of it was measured on you.

What is a good ApoB level?

No trial has established an ApoB target, which is the honest answer rather than a dodge. Europe is the only body that prints ApoB numbers as goals: under 65, 80 and 100 mg/dL for very-high, high and moderate-risk people. It calls them secondary, because the primary goal there is still LDL cholesterol, and a risk group is something your physician assigns you rather than something you pick. Canada declines to set one, and says why [13]. To date, no clear target to which LDL cholesterol, non-HDL cholesterol or ApoB levels should be lowered is clearly identified in randomized trials. The American guideline names a level at or above 130 mg/dL as a risk-enhancing factor [12]. It attaches no class of recommendation and no level of evidence, so that is a flag for a conversation rather than a goal. Read your result against your own laboratory's reference range and its units.

Should I ask my doctor for an ApoB test?

The clearest case, and close to the European guideline's own wording, is if you have high triglycerides, diabetes, obesity, metabolic syndrome or a very low LDL cholesterol [11]. Those are the situations where the two numbers are most likely to disagree. It runs on blood you are already giving, so it does not mean another appointment. It is not a replacement for your panel and it is not something to chase downward.

Does a normal cholesterol panel mean my lipids are fine?

Not necessarily, and for a reason specific to this panel. The lipid bands on a laboratory report are not a reference interval drawn from a healthy population the way most of your other markers are. They are clinical decision limits, thresholds chosen to flag a risk, which is a distinction I unpack in the bloodwork piece. So a lipid result inside the printed range tells you something different from a sodium result inside its range. ApoB is one of the ways that difference becomes visible.

Is ApoB better than LDL cholesterol?

The evidence leans that way and stops short of settling it. A 2011 meta-analysis of 12 reports in 233,455 people found ApoB a stronger predictor than both non-HDL and LDL cholesterol [9]. Its lead author has argued the case for ApoB in print for two decades, and its funding statement is unreadable. A genetic study across roughly 440,000 UK adults found ApoB the predominant trait when all three were modeled together [8]. But the genetic study's own interval for LDL cholesterol ran from 0.57 to 1.27, meaning the model could not cleanly separate them, and its authors say so. Guidelines reflect that uncertainty: Canada prefers ApoB in some patients, Europe allows it, America lists it, and the UK does not mention it.

Supporting Evidence

1. Wilkins, J. T., Li, R. C., Sniderman, A., Chan, C., & Lloyd-Jones, D. M. (2016). Discordance Between Apolipoprotein B and LDL-Cholesterol in Young Adults Predicts Coronary Artery Calcification: The CARDIA Study. Journal of the American College of Cardiology, 67(2), 193-201. 2,794 participants followed 25 years, prospective observational cohort. NHLBI contracts; authors declare no relevant relationships. Allan D. Sniderman, who has argued the case for ApoB in print for two decades, is a co-author. Read 1 October 2026. doi:10.1016/j.jacc.2015.10.055
2. Kim, C.-W., Hong, S., Chang, Y., Lee, J. A., Shin, H., & Ryu, S. (2021). Discordance Between Apolipoprotein B and Low-Density Lipoprotein Cholesterol and Progression of Coronary Artery Calcification in Middle Age. Circulation Journal, 85(6), 900-907. Median follow-up 2.6 years, the shortest of the cohorts cited here. 14,205 Korean men, prospective observational cohort. The paper carries no funding statement at all; conflicts stated as "Nothing to report". Read 1 October 2026. doi:10.1253/circj.cj-20-0692
3. Giannakopoulou, S.-P., et al. (2025). Discordance between apolipoprotein B and LDL-cholesterol and 20-year cardiovascular disease risk: the ATTICA study. European Journal of Clinical Investigation. The 2,208 in the ApoB/LDL threshold table is a count summed from that table's cells, not a figure the paper prints. Its analysed follow-up sample is 1,988 subjects, which is what the hazard ratios are computed on. Followed 20 years. Funded by Hellenic Cardiology Society and Hellenic Atherosclerosis Society grants; authors declare no conflict. Read 1 October 2026. doi:10.1111/eci.70077
4. Mora, S., Buring, J. E., & Ridker, P. M. (2014). Discordance of Low-Density Lipoprotein (LDL) Cholesterol With Alternative LDL-Related Measures and Future Coronary Events. Circulation, 129(5), 553-561. 27,533 women, median follow-up 17.2 years. Interested party: the first author reports research grant support from AstraZeneca and Atherotec Diagnostics, and consulting for Pfizer, Quest Diagnostics and Genzyme. The paper's own grant funding is NIH. Read 1 October 2026. doi:10.1161/CIRCULATIONAHA.113.005873
5. Solnica, B., et al. (2023). Discordance between LDL-C, non-HDL-C and apoB in a Polish population: the NATPOL 2011 survey. International Journal of Cardiology, 390, 131150. Cross-sectional national survey with no outcomes; read as abstract only. Funding not readable; "Declaration of Competing Interest None declared"; Allan Sniderman is a co-author. Read 1 October 2026. doi:10.1016/j.ijcard.2023.131150
6. Du, Y., et al. (2025). Discordance between apolipoprotein B and LDL cholesterol and cardiovascular outcomes. European Journal of Preventive Cardiology. 375,544 UK Biobank adults with no baseline cardiovascular disease and none on lipid-lowering therapy. The paper's own funding and competing-interest statements could not be read; the publisher returned 403. The Europe PMC grant record names Chinese Academy of Medical Sciences and NSFC awards, which is index metadata rather than the paper's words. Read 1 October 2026. doi:10.1093/eurjpc/zwaf750
7. Cao, J., et al. (2020). Apolipoprotein B discordance and coronary artery calcium in MESA. Journal of Clinical Lipidology, 14(1). 4,623 statin-free adults aged 45 to 84. NHLBI contracts; "DECLARATIONS OF INTEREST: None". The open copy is the unedited accepted manuscript. Read 1 October 2026. doi:10.1016/j.jacl.2019.11.005
8. Richardson, T. G., et al. (2020). Evaluating the relationship between circulating lipoprotein lipids and apolipoproteins with risk of coronary heart disease: A multivariable Mendelian randomisation analysis. PLOS Medicine, 17(5), e1003062. Genome-wide association study plus multivariable Mendelian randomisation; genetic epidemiology on observational data, so association rather than trial evidence. Bristol MRC unit funding. One co-author reports grants and advisory fees from most of the large lipid-drug companies, and is a named task-force member on the European guideline cited here. Read 1 October 2026. doi:10.1371/journal.pmed.1003062
9. Sniderman, A. D., et al. (2011). A Meta-Analysis of Low-Density Lipoprotein Cholesterol, Non-High-Density Lipoprotein Cholesterol, and Apolipoprotein B as Markers of Cardiovascular Risk. Circulation: Cardiovascular Quality and Outcomes, 4(3), 337-345. 12 reports, 233,455 subjects, 22,950 events. Funding and conflict statements are genuinely unreadable by every route tried: no open copy, publisher 403, repository holds metadata only. The label owed the reader is that the lead author has argued the case for ApoB in print for two decades, from his published record. Read 1 October 2026. doi:10.1161/CIRCOUTCOMES.110.959247
10. Marston, N. A., et al. (2022). Association of Apolipoprotein B-Containing Lipoproteins and Risk of Myocardial Infarction in Individuals With and Without Atherosclerosis. JAMA Cardiology, 7(3), 250-256. 389,529 primary-prevention adults on no lipid-lowering therapy, median follow-up 11.1 years. FOURIER was funded by Amgen and IMPROVE-IT by Merck; no funding was provided for this analysis; the authors carry extensive industry grants. The paper does not print the standard deviation, so its figure cannot be converted to mg/dL. Read 1 October 2026. doi:10.1001/jamacardio.2021.5083
11. Mach, F., Baigent, C., Catapano, A. L., et al. (2020). 2019 ESC/EAS Guidelines for the management of dyslipidaemias. European Heart Journal, 41(1), 111-188. The ApoB recommendation carries Class I, Level C, which the same document defines as consensus of expert opinion, small studies, retrospective studies or registries. The 2025 focused update does not change the ApoB position. Read 1 October 2026. doi:10.1093/eurheartj/ehz455
12. Grundy, S. M., Stone, N. J., et al. (2019). 2018 AHA/ACC/AACVPR/AAPA/ABC/ACPM/ADA/AGS/APhA/ASPC/NLA/PCNA Guideline on the Management of Blood Cholesterol. Circulation, 139(25), e1082-e1143. ApoB appears in Table 6 among risk-enhancing factors, under "If measured". It carries no class of recommendation and no level of evidence anywhere in the version I could read; the publisher's typeset copy returned 403. Read 1 October 2026. doi:10.1161/CIR.0000000000000625
13. Pearson, G. J., Thanassoulis, G., Anderson, T. J., et al. (2021). 2021 Canadian Cardiovascular Society Guidelines for the Management of Dyslipidemia for the Prevention of Cardiovascular Disease in Adults. Canadian Journal of Cardiology, 37(8), 1129-1150. The screening recommendation is graded a strong recommendation on high-quality evidence. Provenance note: the chapter text was read from the society's own site, which serves an incomplete certificate chain. The same recommendation was corroborated independently from the publisher-supplied MEDLINE abstract. Read 1 October 2026. doi:10.1016/j.cjca.2021.03.016
14. National Institute for Health and Care Excellence (2023). Cardiovascular disease: risk assessment and reduction, including lipid modification. NICE guideline NG238, published 14 December 2023. Every published chapter was searched: apolipoprotein and apob occur zero times; non-HDL cholesterol is used instead. Read 1 October 2026. www.nice.org.uk/guidance/ng238