Lp(a) test: who should get one, cost and preparation

Labs measure Lp(a) with antibodies against apo(a), the protein that defines the particle. Because apo(a) varies in size, the best assays are built so that large and small forms read the same, and they report particle counts in nmol/L [8]. Guidelines prefer nmol/L but accept mg/dL [3].

The test is separate from LDL cholesterol and ApoB. Many people order both ApoB and Lp(a) together: ApoB counts every artery-clogging particle, while Lp(a) shows how many carry the extra apo(a) protein. Testing remains uncommon: in the University of California health system, only 1.8% of adults who had a lipid panel also had an Lp(a) test [9].

Universal testing is now the majority view. The EAS (2022), the National Lipid Association (2024), the ESC/EAS (2025) and the ACC/AHA (2026) all advise at least one test in every adult [1][2][3][4]. Canada has recommended one test with the first lipid screen since 2021 [10].

HEART UK recommends testing these groups [11]:

• people with early heart disease, or a family history of it
• first-degree relatives of someone with Lp(a) above 200 nmol/L
• people with familial hypercholesterolemia
• people with calcific aortic valve stenosis
• people with a borderline 10-year risk (below 15%)

Children and teenagers

The EAS suggests testing young people with a history of ischemic stroke, or whose parent had early heart disease without another explanation. Levels can still rise until adulthood, so a childhood result may need repeating [1].

Relatives

In UK Biobank, 47% of first-degree relatives and 32% of second-degree relatives of someone with high Lp(a) also had high Lp(a), against 16% of unrelated pairs [7]. One high result is a reason to test parents, siblings and children.

In 34,829 Danish adults, median Lp(a) was 17.3 mg/dL in fasting samples and 19.4 mg/dL three to four hours after a meal, a difference that was not statistically significant. Lp(a) predicted heart disease equally well either way [5]. The EAS lists fasting as having no effect on Lp(a) [1].

If the same draw includes a full lipid panel, nonfasting samples are also accepted for most other lipids. Triglycerides are the main exception; see triglycerides.

Lp(a) is highly consistent within a person over years [12], and the 2026 ACC/AHA guideline summary notes that one early measurement is enough for most people [4]. The EAS says repeating does not improve risk prediction for most people, with exceptions such as kidney or liver disease and acute infection [1].

Repeat tests are not identical. In 250 lipid clinic patients, results varied by about 19% within a person, and 12.4% moved across the NLA cutoffs of 75 or 125 nmol/L on retesting [6]. The ESC/EAS 2025 update adds that levels may rise after menopause, so a second test is reasonable if a premenopausal result was borderline [3].

We do not quote a price because lab and insurer prices change often and differ widely. A few practical points:

• Ask for it by name. It is not part of a routine cholesterol panel, so the order needs to say lipoprotein(a) or Lp(a).
• Ask about coverage. Coverage often depends on whether there is a medical reason recorded, such as family history of early heart disease, which every guideline lists as a reason to test [11][1].
• Check the unit. If you have a choice, a lab that reports nmol/L with an assay that reads all apo(a) sizes evenly is preferred [8][3].

These cutoffs come from the EAS consensus and the NLA update [1][2]. Risk rises steadily rather than jumping at one line: in UK Biobank it rose 11% for every 50 nmol/L [13], and the ACC/AHA summary links above 250 nmol/L to at least double the risk [4].

A result in mg/dL cannot be converted exactly into nmol/L, because the weight of each particle depends on apo(a) size [8]. The full range chart and a guideline comparison are on what is a good Lp(a) level, and next steps for a high result are on what a high Lp(a) means.

Apo(a) size varies between people and between a person's two gene copies, so antibodies may bind a different number of sites on each particle, and results from different labs may not agree [8]. This is why the ESC/EAS prefers nmol/L [3], and Japan's 2026 consensus calls for assays aligned with international (IFCC) standards [14].

Normal biological variation adds about 19% from one test to the next [6]. If your result sits close to a cutoff, a repeat test at the same lab gives a clearer picture.