Lipoprotein(a) — Lp(a)
An LDL-like particle with an added apolipoprotein(a) tail. Your level is roughly 90% genetically determined and essentially fixed for life.
An independent, causal risk factor for heart attack, stroke and aortic stenosis that almost nobody tests. Because it's genetic and stable, it only needs to be measured once in a lifetime — and every current guideline says to do exactly that.
Check a Lipoprotein(a) — Lp(a) result against this range →
What Lipoprotein(a) — Lp(a) actually measures — the analyte, and the assay
The analyte is a particle, not a molecule: an LDL-like core carrying apolipoprotein B, with a second protein — apolipoprotein(a) — welded to it by a disulfide bond. Apo(a) is a plasminogen relative built from looped kringle domains, and one of those domains, kringle IV type 2, is present in a number of copies that varies from a handful to several dozen between people. Apo(a) therefore has no fixed size, ranging across hundreds of kilodaltons, and the number of repeats you inherited is inversely related to how much Lp(a) you circulate Volgman 2024.
That variable protein is why this test has two units that are not interchangeable. A mass assay reports mg/dL and measures the total protein present; if its antibody binds an epitope inside the repeated kringle, a person with a large isoform generates more signal per particle and reads high, while a small-isoform carrier reads low. A molar assay reports nmol/L and is designed to count particles regardless of size, which is what the risk actually tracks. Building an assay that genuinely does not care about isoform size is difficult enough to be a publishable achievement in its own right: an isoform-independent monoclonal ELISA validated over an analytical measuring range of 0.27–1,402 nmol/L, correlating at r = 0.987 and r = 0.976 with comparison methods Marcovina 2022.
There is no valid conversion factor between the two units, and this is where real results get misread. In 11,970 participants of a cardiovascular outcome trial, the same samples gave a median of 21.8 mg/dL on one test and 45.0 and 42.2 nmol/L on two molar tests Szarek 2023. That is a ratio near 2.1 in that cohort — not the 2.5 that circulates as a rule of thumb, and in any case a ratio that changes person by person with isoform size. Multiply your mg/dL by 2.5 and you can move yourself across a treatment threshold that nothing in your blood has crossed. The practical guidance is to ask for the molar result and never to convert Cegla 2021.
Within a single unit the assays are precise. Five commercial Lp(a) immunoassays run on one chemistry analyzer gave coefficients of variation of 2.5–5.2% on a low pool averaging 7.3–12.4 mg/dL and 0.8–3.0% on a high pool averaging 31.5–50.2 mg/dL Wyness 2021. Note what those numbers say: the five assays each measure reproducibly, and they do not agree with each other about the concentration — the low pool alone spanned 7.3 to 12.4 mg/dL depending on which one ran it.
Lipoprotein(a) — Lp(a): what changes the blood, and what only changes the reading
What changes the Lp(a) in your blood — and this is the shortest such list in the estate:
- The LPA gene you inherited. Between 80% and 90% of the between-person variation is genetic, set by kringle repeat number and by variants in and around the locus Volgman 2024 Kronenberg 2022. Nothing else on this list comes close.
- Ancestry, which is the population-level expression of the same genetics: median concentrations differ substantially between ancestral groups, so a threshold derived in one population is not automatically the right threshold in another Volgman 2024.
- Kidney function. Chronic kidney disease and especially nephrotic syndrome raise it, and the rise reverses if the kidney problem does.
- Thyroid status. Untreated hypothyroidism raises Lp(a) and replacement lowers it — one more reason the thyroid panel and the lipid panel are read together rather than separately.
- Estrogen. Concentrations rise after menopause and fall on estrogen therapy.
- Inflammation. Lp(a) behaves as a modest acute-phase reactant, so a value drawn in the weeks after a heart attack, an operation or a serious infection is not the baseline.
- Drugs. PCSK9 inhibitors and niacin lower it modestly; the RNA-targeted agents built specifically against apo(a) lower it far more and were still in outcome trials when the European consensus was written Kronenberg 2022. Statins do not lower it.
What changes only the reading — and here the list is longer than the biology:
- Mass versus molar reporting. The dominant term, and the reason two results a year apart can look like a change that never happened Szarek 2023.
- Isoform-dependent bias in any assay whose antibody sees the repeated kringle Marcovina 2022.
- Which of several commercial assays the laboratory bought Wyness 2021.
- A conversion applied by a website, an app or a clinician working from memory. This is not a laboratory error, and it is probably the commonest way a person ends up with the wrong Lp(a) in their head Cegla 2021.
Reference interval or decision threshold — which kind of number Lipoprotein(a) — Lp(a) is
This is a risk threshold pretending to be a reference interval, and the pretense does more harm here than anywhere else on the site. Lp(a) is not distributed like a normal laboratory analyte: the population distribution is steeply right-skewed, with a large fraction of people near the bottom of the scale and a long tail. A 'central 95%' of that shape is not a meaningful statement, and no laboratory's reference interval is doing the work people think it is.
What exists instead are concentrations above which measured event rates rise, derived from cohorts and genetic studies and set out in the European Atherosclerosis Society's consensus statement Kronenberg 2022. Those are decision thresholds in the strict sense, and — the point of the assay section above — they are quoted in molar units because that is the unit the risk was measured in. A threshold expressed in nmol/L cannot be applied to a result expressed in mg/dL without an isoform-dependent error Cegla 2021 Szarek 2023.
Risk is also continuous rather than switched. There is no concentration at which nothing happens below and something happens above; the thresholds are places on a slope chosen to be clinically actionable. Recent work has begun applying the same logic outside coronary disease, assessing apoB and Lp(a) for sex-specific aortic stenosis risk Shi 2026 — a disease with no lipid-lowering treatment at all, which is a useful reminder that a risk threshold is not automatically a treatment threshold.
How you would know your Lipoprotein(a) — Lp(a) was wrong — and when to redraw
Once, ever, for most people — and the reason is genetic rather than practical. Concentration is 80–90% determined by a locus you were born with Volgman 2024, it is stable from childhood, and it does not respond to diet, exercise or weight change. The consensus position is that a single measurement in a lifetime is enough to classify most adults Kronenberg 2022. That makes it the cheapest permanent piece of information on the whole panel.
Five circumstances genuinely justify a second measurement, and only these: the first result was reported in mg/dL and a molar assay is now available; it was drawn within weeks of a myocardial infarction, surgery or serious infection, when Lp(a) is transiently raised; new or worsening kidney disease; a change in thyroid status or the start of treatment for it; or starting, stopping or changing estrogen therapy.
What would have to change for the second value to mean anything. Given the genetics, a real change of any size demands an acquired explanation from that list. Everything else is method. Before believing that your Lp(a) has moved, confirm that both results are in the same unit, from the same assay, in the same clinical state — and if any of the three differs, the change is in the measurement Szarek 2023 Wyness 2021.
How you would know the value was wrong. An Lp(a) that 'improved' after a diet, a supplement or a period of training is the clearest possible signal of an assay or unit change, because none of those interventions has been shown to move it. The confirmatory step is to obtain the original report, read the unit, and ask which assay each laboratory ran Cegla 2021.
What Lipoprotein(a) — Lp(a) cannot tell you
It cannot tell you when, only whether. Lp(a) describes a lifelong exposure. It does not predict an event in a window, and a high value in a 30-year-old is a reason to control everything else aggressively for decades, not a forecast.
A low Lp(a) is not a clean bill of health. It removes one independent contributor and leaves apoB, blood pressure, glycemia and smoking exactly where they were.
It is already inside your LDL-C, and that distorts both numbers. Lp(a) particles carry cholesterol which the laboratory counts as LDL cholesterol. When Lp(a)-cholesterol was quantified directly it ranged from 0.6 to 35.0 mg/dL and accounted for anywhere between 5.8% and 57.3% of measured LDL-C; correcting for it moved mean LDL-C from 119.2 mg/dL to 102.2 mg/dL Yeang 2021. In someone with a high Lp(a), part of the 'LDL cholesterol' being treated belongs to a particle that statins barely touch.
Results in different units are not comparable and must not be trended — the single most consequential limitation on this page Szarek 2023.
The wrong inference readers actually draw is that a high Lp(a) is something to fix. As of the European consensus, the therapies capable of lowering it substantially were still in outcome trials and no lipid intervention had been shown to reduce events by lowering Lp(a) specifically Kronenberg 2022. What a high result changes is the aggressiveness with which everything modifiable is treated — and that is a real change, made on information you only have to buy once.
Sources read for these sections
- Kronenberg F, et al. Lipoprotein(a) in atherosclerotic cardiovascular disease and aortic stenosis. European Heart Journal 2022 · PMID 36036785
- Cegla J, et al. Lp(a): When and how to measure it. Annals of Clinical Biochemistry 2021 · PMID 33040574
- Szarek M, et al. Relating Lipoprotein(a) Concentrations to Cardiovascular Event Risk After Acute Coronary Syndrome: A Comparison of 3 Tests. Circulation 2023 · PMID 37632469
- Marcovina SM, et al. Development and validation of an isoform-independent monoclonal antibody-based ELISA for measurement of lipoprotein(a). Journal of Lipid Research 2022 · PMID 35688187
- Wyness SP, et al. Performance evaluation of five lipoprotein(a) immunoassays on the Roche cobas c501 chemistry analyzer. Practical Laboratory Medicine 2021 · PMID 33898688
- Yeang C, et al. Novel method for quantification of lipoprotein(a)-cholesterol: implications for improving accuracy of LDL-C measurements. Journal of Lipid Research 2021 · PMID 33636163
- Volgman AS, et al. Genetics and Pathophysiological Mechanisms of Lipoprotein(a)-Associated Cardiovascular Risk. Journal of the American Heart Association 2024 · PMID 38879448
- Shi W, et al. Moving beyond Low-Density Lipoprotein cholesterol: apolipoprotein B and Lipoprotein (a) for sex-specific risk assessment of aortic stenosis in the UK Biobank. European Journal of Preventive Cardiology 2026 · PMID 42104638
The plan of attack
In this order. Most people start at step four, which is why they change five things at once and learn nothing.
- Confirm the number is real
Units — mg/dL vs nmol/L. Not a biological interference but the most common misreading of this marker. The two units are not interchangeable and there is no exact conversion, because the assay depends on particle size. Use nmol/L where possible, and never compare a mg/dL result to an nmol/L threshold. - Read it with its partner
Test once, ever. If elevated, it permanently reframes how hard you should treat ApoB and blood pressure — one of the highest-value single tests in all of medicine, and it costs $25. Draw it alongside: ApoB (Apolipoprotein B), Lipid Panel (Cholesterol, HDL, LDL, Triglycerides), hs-CRP (High-Sensitivity C-Reactive Protein). - Work out which direction is yours
If it's high — Means every other risk factor — ApoB, blood pressure, insulin resistance, smoking — must be controlled more aggressively, because your baseline risk is already elevated. It also strengthens the case for early imaging (coronary calcium score). - Fix it in this order
Nutrition. Diet barely moves Lp(a) — and that's not a failure, it's the biology. Don't chase it with food; control ApoB instead.
Lifestyle. Since you can't lower it much, the play is aggressively managing everything you can control — blood pressure, ApoB, glucose, smoking, fitness.
Supplements. Niacin lowers it modestly (~20%) but hasn't shown outcome benefit and has side effects. No supplement meaningfully helps.
Hormones. PCSK9 inhibitors lower Lp(a) ~25%. Targeted RNA therapies (pelacarsen, olpasiran) are in phase 3 and may change this entirely. Estrogen therapy lowers it in postmenopausal women.
Compounds. No peptide interaction. Purely a risk-stratification test.
Work down the list, not across it. Adding a compound on top of an unfixed diet is why generic protocols fail. - Retest
Once in a lifetime is sufficient unless kidney or thyroid status changes markedly. Change one thing at a time, or the retest can't tell you which thing worked.
How to fix it
📚 2026 AHA/ACC Dyslipidemia Guideline — measure at least once. Tsimikas S, J Am Coll Cardiol 2017 — Lp(a) as a causal risk factor.
This page can tell you what could have made your Lipoprotein(a) — Lp(a) wrong. It cannot tell you whether it did.
Everything above is free and stays free — the assay, what changes the reading rather than the blood, the retest window and the sources. What no page can do is look at your draw: which laboratory ran it, at what hour, what you were taking that week, and what else was flagged beside it. Every one of those changes the answer, and none of them is on any page. Bringing a real result to people who know that list is what the members' area is for.
Bring your result — $10/mo →🩸 Test your Lipoprotein(a) — Lp(a)
Order directly through Marek Diagnostics — no doctor's visit needed, drawn at any Quest location in the US. Code CAMERON applies 10% off automatically.
Order this test — 10% off → Browse all 103 markers →What Lipoprotein(a) — Lp(a) is usually tested alongside
One marker is a data point. These panels add the markers that make Lipoprotein(a) — Lp(a) interpretable, name why each is on the list, and load the set into your cart at 10% off.
includes this + 6 more markers — Taking atorvastatin, rosuvastatin, simvastatin or any other statin, or about to start one. Also for anyone who stopped a statin because of muscle aches and wants to know what was actually happening.
includes this + 9 more markers — Anyone eating ketogenic, carnivore or seriously low-carbohydrate — especially anyone whose LDL cholesterol went up sharply after starting, and anyone who feels cold, flat or is losing training performance months in.
What people use Lipoprotein(a) — Lp(a) to decide
Nobody orders a test for its own sake. Lipoprotein(a) — Lp(a) is on the test list for these pathways — each one links to what the pathway claims, and what its test list is read for before you spend anything on it.
ApoB counts atherogenic particles; LDL-C estimates the cholesterol inside them. When the two disagree — common in insulin resistance — ApoB is right and LDL-C is falsely reassuring.
These are the numbers with real causal mortality evidence. ApoB beats LDL-C, Lp(a) is genetic and worth measuring exactly once in your life, and cystatin-C catches kidney decline that creatinine misses. If you test nothing else on this page, test these.
Measure Lp(a) once in your life. It is genetic, unmoved by diet or exercise, raised in about one in five people, and almost never ordered — and it changes how aggressively everything else on this page should be treated.
Would you feel it? Symptoms Lipoprotein(a) — Lp(a) helps explain
People search for how they feel, not for a marker. These are the complaints where this one is worth checking, and whether it is first-line or a follow-up.
Why your Lipoprotein(a) — Lp(a) might be wrong
Most abnormal results are interference, not disease. Check these before you change anything. Each says whether the number is wrong (repeat it), badly timed (redraw it), or real with a cause.
Lp(a) is a modest acute-phase reactant and rises during inflammation, which is the main reason a repeat value differs from the first.
Measure when well. And you only need it once — it is genetically fixed.
Not a biological interference but the most common misreading of this marker. The two units are not interchangeable and there is no exact conversion, because the assay depends on particle size.
Use nmol/L where possible, and never compare a mg/dL result to an nmol/L threshold.
What Lipoprotein(a) — Lp(a) means in combination
One marker tells you a little; combinations tell you the story. These are the named patterns this one takes part in.
Lp(a) is almost entirely genetic, largely unmoved by diet, and independently raises cardiovascular and aortic valve risk. A normal LDL alongside it is falsely reassuring.
Measure it once — it barely changes over a lifetime. You cannot lower it much yet, so the response is to drive everything else down hard: ApoB, blood pressure, insulin resistance, smoking. Tell your first-degree relatives to test theirs.
Several independent contributors to clot risk stacking together. Each is unremarkable alone; the combination is not, particularly alongside androgen use, high hematocrit, estrogen therapy or long-haul travel.
A raised D-dimer with leg swelling, chest pain or breathlessness is an emergency, not a retest. Otherwise: address homocysteine with B12/folate/B6, treat the inflammation driving fibrinogen, and review anything estrogenic or androgenic you are taking with a clinician.
What to test next
These put Lipoprotein(a) — Lp(a) in context — each with its own full breakdown.
Frequently asked questions
<75 nmol/L (or <30 mg/dL) reported as normal — risk is continuous. Ranges vary by laboratory and assay — always compare to the range printed on your own report.
<75 nmol/L desirable. ≥125 nmol/L (~50 mg/dL) is a recognized risk-enhancing threshold (~1.4× ASCVD risk); 300–400 nmol/L approaches heterozygous familial hypercholesterolemia risk.
Means every other risk factor — ApoB, blood pressure, insulin resistance, smoking — must be controlled more aggressively, because your baseline risk is already elevated. It also strengthens the case for early imaging (coronary calcium score).
Favorable; nothing to act on.
You can order Lipoprotein(a) — Lp(a) directly through Marek Diagnostics without a doctor's visit — drawn at any Quest Diagnostics location in the US. Code CAMERON applies 10% off automatically.
Where this goes next
This page is the free framework. The protocol itself — the dosing, the order to correct things in, the week-by-week schedule and what to retest — is a lesson inside Skool.