Lp(a) Levels: nmol/L vs mg/dL, Risk Cutoffs, and What Actually Lowers It

An Lp(a) below 75 nmol/L (roughly 30 mg/dL) is low risk, 75–125 nmol/L (about 30–50 mg/dL) is a borderline grey zone, and above 125 nmol/L (about 50 mg/dL) is elevated — the level at which the 2022 European Atherosclerosis Society (EAS) consensus statement and most laboratories flag clearly increased risk of heart attack, stroke and aortic valve stenosis. Roughly one in five people worldwide sits above that line. Lipoprotein(a) is unusual among lipid numbers in two ways: your level is more than 90 percent genetically determined, so diet and exercise barely move it, and labs report it in two different units — nmol/L (particle count) and mg/dL (particle mass) — that measure different things and cannot be exactly converted into each other.

The honest note on units: nmol/L counts Lp(a) particles; mg/dL weighs their total mass. Because the apolipoprotein(a) protein varies enormously in size between people, there is no exact conversion factor. The rule of thumb — multiply mg/dL by about 2.15 to approximate nmol/L — is a population average that can be off substantially for any individual and depends on the assay. The two columns above are parallel scales taken from the same guidance, not converted values. Never convert your own result; if you need the other unit, retest in it.

What actually moves Lp(a)?

Almost nothing you can buy in a supplement store, and nothing on a treadmill. Here is the honest table:

That last column is the uncomfortable part of the Lp(a) story in 2026, and we come back to it below. First: what your own number means.

My Lp(a) is 90 nmol/L — what does that mean?

90 nmol/L lands in the grey zone: above the 75 nmol/L line where Lp(a) starts adding meaningful risk, below the 125 nmol/L cutoff labs flag as elevated. The population distribution is heavily skewed — the median sits far below 75 nmol/L — so 90 is well above typical, and it will not change, because Lp(a) is set by your LPA gene and stays roughly stable for life. Risk is continuous, not a cliff: the EAS models it rising smoothly with both level and years of exposure. At 90, the added risk is real but modest.

First, check the unit — this is where people get burned. 90 nmol/L is borderline; 90 mg/dL is a completely different result (roughly 190–200 nmol/L on the rule of thumb) and clearly elevated. Reading your lab report with the wrong unit in mind moves you two risk bands.

What a 90 nmol/L should change is arithmetic, not anxiety: your cardiovascular risk is Lp(a) multiplied by everything else. A 90 with an ApoB of 70 mg/dL, normal blood pressure and no smoking is a footnote. The same 90 stacked on an ApoB of 120, borderline hypertension and a family history of early heart disease is a genuine argument for treating those other numbers aggressively. Log the result once in BioTrakk — with its unit — and it sits permanently next to your ApoB trend, where it belongs.

How do you convert Lp(a) from mg/dL to nmol/L?

You do not — not exactly, and every serious source says so. The reason is the apolipoprotein(a) protein itself: it contains a repeating domain (kringle IV type 2) whose copy number varies from a handful to more than 40, so the protein can differ several-fold in size between two people. Two people with the same particle count (nmol/L) can have very different particle mass (mg/dL), and vice versa. Mass assays are also calibrated differently between manufacturers, which is why Tsimikas’ JACC review and the EAS consensus both recommend isoform-insensitive assays reporting in nmol/L.

The practical rules:

• 1 mg/dL ≈ 2.15 nmol/L is a rough population average — fine for orienting yourself, wrong for precise classification near a cutoff.
• If your result is within about 20 percent of a decision threshold in converted units, do not classify yourself from the conversion. Retest in nmol/L.
• Track one unit, forever. A switch of lab or assay can move your apparent Lp(a) without anything changing in your blood.

Why doesn’t lifestyle lower Lp(a)?

Because Lp(a) is not a metabolic readout — it is a genetic setting. More than 90 percent of the variation between people is explained by the LPA gene, and your level is largely fixed by early adulthood. Diet changes that reliably move LDL and ApoB do essentially nothing to Lp(a); if anything, trials that cut saturated fat have sometimes seen Lp(a) tick slightly upward while LDL falls. Exercise does not lower it. Weight loss does not lower it. No supplement has credible evidence for lowering it meaningfully. Anyone selling you an Lp(a)-lowering protocol of berries, saunas or fish oil is selling you noise.

Niacin is the cautionary tale: it genuinely lowers Lp(a) by 20–30 percent, which made it the great hope for years. Then AIM-HIGH and HPS2-THRIVE — over 29,000 patients between them — found no reduction in cardiovascular events, and HPS2-THRIVE found significant harms (new-onset diabetes, bleeding, infections). Lowering a number is not the same as lowering risk. No guideline recommends niacin for Lp(a) today.

This page will not pretend otherwise, because the honest version is genuinely useful: it frees you from blaming your diet for a number your diet never controlled, and points your effort at the risk factors that do respond.

My Lp(a) is high — what should I actually do?

Four things, in order:

1. Confirm it once, then stop retesting. One good measurement in nmol/L is enough for life for most people, because the level barely changes. Quarterly Lp(a) panels are wasted money.
2. Attack the risk factors that are modifiable. This is the entire actionable strategy in 2026, and it works: get ApoB low (high Lp(a) is a recognized argument for earlier and more aggressive lipid-lowering), control blood pressure, do not smoke, keep glucose and insulin in range. In the FOURIER trial, patients with higher Lp(a) got the largest absolute benefit from PCSK9 inhibition — evolocumab lowered Lp(a) by about 25 percent alongside LDL. Whether a PCSK9 inhibitor is appropriate for you is a medical decision, but it is a conversation worth having if your Lp(a) is high and your ApoB is not at goal.
3. Tell your family. Lp(a) is inherited roughly co-dominantly; each first-degree relative has about a 50 percent chance of sharing an elevated level. The EAS explicitly recommends cascade testing of relatives when very high Lp(a) is found. One cheap blood test per relative, once.
4. Keep the rest of your panel honest. Lp(a) travels with other quietly genetic markers worth knowing once — homocysteine and ferritin among them — and none of them mean much in isolation. Risk lives in the combination.

BioTrakk is not a medical device and this is not medical advice — but walking into a cardiology appointment with your Lp(a), a dated ApoB trend and your blood pressure log makes that appointment dramatically better.

What is honestly happening with the new Lp(a) drugs?

The field just had its first big test — and failed it. On 4 September 2026, Novartis announced that Lp(a)HORIZON — the phase 3 outcomes trial of pelacarsen, an antisense drug that lowers Lp(a) by roughly 80 percent, in 8,323 patients with established cardiovascular disease and Lp(a) of at least 70 mg/dL — did not meet its primary endpoint. The drug lowered Lp(a) as designed; cardiovascular events did not fall versus placebo. Full data are due at an upcoming congress.

What that means, stated carefully: lowering Lp(a) by 80 percent with this drug, in this population, on top of modern statin-era care, did not reduce events. It does not yet prove the whole hypothesis dead. The genetic evidence that lifelong high Lp(a) causes heart disease remains strong, and deeper lowering may matter: olpasiran, an siRNA that suppressed Lp(a) by 95–100 percent in OCEAN(a)-DOSE, has its ~7,000-patient outcomes trial (OCEAN(a)-Outcomes, in patients with Lp(a) of at least 200 nmol/L) estimated to complete around the end of 2026. A third agent, lepodisiran, is in an even larger phase 3 that runs to late this decade. Within a year or two we will know whether pelacarsen was the wrong molecule, the wrong population — or whether Lp(a) lowering in adulthood simply comes too late.

Until an outcomes trial is positive, the honest position is: there is no approved drug for lowering Lp(a), and no proof yet that lowering it in adulthood prevents events. Which is exactly why the previous section — managing the modifiable risks — is the strategy, not the consolation prize.

Who should test Lp(a) — and how often?

Everyone, once. That is the formal recommendation of the EAS 2022 consensus: every adult should have Lp(a) measured at least once in their lifetime, because one in five people carries a clinically relevant elevation, it is invisible on a standard cholesterol panel, and it reclassifies risk in ways that change treatment intensity. It is a cheap add-on to any lipid draw. Priority goes to anyone with a personal or family history of early cardiovascular disease, familial hypercholesterolaemia, or relatives with known high Lp(a).

Once is genuinely enough for most people. Reasonable exceptions for a second measurement: your result sits close to a decision threshold (assay variation is real), a major hormonal transition such as menopause, significant kidney disease (which raises Lp(a)), or your lab switched assay or units. Outside those, retesting Lp(a) every quarter tells you nothing new — put that budget toward the markers that actually move, and track them properly. Log your one Lp(a) result in BioTrakk by chat, in nmol/L, and let it do its job: quietly recalibrating how hard you chase everything else.