Lp(a): Test It Once, Then Watch the Drug Pipeline

The full story

There is a blood test you can reasonably run once, look at, and never repeat. The American Heart Association recommends that every adult have lipoprotein(a) measured at least once in a lifetime; the National Lipid Association says the same thing in its 2024 focused update. Meanwhile the National Heart, Lung, and Blood Institute notes, drily, that an Lp(a) test "is not usually part of a routine lipid panel." Elevated Lp(a) affects roughly one in five people, and measurement rates remain low.

The strangeness of this test is that it sits at an awkward intersection: a risk factor with unusually good causal credentials, published thresholds, and - at the point the phase 3 protocols were being written - no approved drug that lowers it. So the obvious objection writes itself. Why measure something you cannot change? The answer is that the number changes what you do about everything else.

The causal case

Most cardiovascular risk factors are known through association. Lp(a) is known through its gene. In the Copenhagen City Heart Study, published in JAMA, the multivariable-adjusted hazard ratio for myocardial infarction rose stepwise with Lp(a) level, reaching 2.6 above the 95th percentile compared with below the 22nd. More importantly, the number of kringle IV type-2 repeats in the LPA gene explained 21% to 27% of the variation in plasma Lp(a) - and the genotypes that predicted higher Lp(a) also predicted higher infarction risk. That is the structure of a causal argument rather than a correlation: the gene that sets the level also sets the risk, and genes are not confounded by lifestyle.

The same design has been run for the aortic valve. Across 77,680 Danish participants followed up to 20 years, with 454 diagnoses of aortic valve stenosis, a study in the Journal of the American College of Cardiology found that Lp(a) above 90 mg/dL - past the 95th percentile - carried a hazard ratio of 2.9 against Lp(a) below 5 mg/dL. Mendelian randomisation gave a genetic relative risk of 1.6 per ten-fold higher Lp(a), close to the observational estimate of 1.4. When the genetic and observational numbers land on top of each other, confounding becomes a much harder story to tell.

The 2022 European Atherosclerosis Society consensus statement in the European Heart Journal pulls this together: epidemiologic and genetic studies in hundreds of thousands of people support a causal, continuous association between Lp(a) and cardiovascular outcomes across ethnicities, and elevated Lp(a) remains a risk factor even at very low LDL-C. That last clause is the practical sting. Driving your LDL-C down does not neutralise this.

Nothing on the pharmacy shelf lowers it

Statins nudge it upward. A subject-level meta-analysis of six randomised trials in the European Heart Journal, covering 5256 patients, found the ratio of geometric means for statin versus placebo was 1.11, with mean changes from baseline of plus 8.5% to plus 19.6% on statin against minus 0.4% to minus 2.3% on placebo. In the laboratory, atorvastatin increased LPA messenger RNA and apolipoprotein(a) protein in HepG2 hepatocytes, so the effect has a plausible mechanism rather than being a statistical artefact.

Read that correctly. It is a real effect and a small one, and it is not an argument against statins - guideline management of a high Lp(a) is early, intensive control of the modifiable risk factors, LDL-C lowering first. It is an argument against expecting your Lp(a) to improve because your other numbers did.

The pipeline is unusually convincing and unusually incomplete

Three injectable agents have posted phase 2 results, all of them RNA-directed therapies rather than anything you could pick up at a pharmacy counter.

• Pelacarsen, a hepatocyte-directed antisense oligonucleotide, in a phase 2 dose-ranging trial in the New England Journal of Medicine: 286 patients with established cardiovascular disease and Lp(a) of at least 60 mg/dL (150 nmol/L), with dose-dependent mean reductions from 35% to 80% at six months, against 6% with placebo. Injection-site reactions were the most common adverse event.
• Olpasiran, a small interfering RNA, in OCEAN(a)-DOSE, also in the New England Journal of Medicine: 281 patients with a median baseline Lp(a) of 260.3 nmol/L. At 36 weeks the placebo-adjusted mean changes ranged from minus 70.5% in the lowest-dose group to minus 97.4%, minus 101.1% and minus 100.5% in the higher-dose regimens, with adverse event rates similar across groups.
• Lepodisiran, a longer-duration siRNA, in the phase 2 ALPACA trial in the New England Journal of Medicine: 320 participants with a median baseline of 253.9 nmol/L. The placebo-adjusted time-averaged reduction from day 60 to day 180 was 40.8 percentage points at the lowest dose, 75.2 at the middle dose and 93.9 in the pooled highest-dose groups. No serious adverse events were attributed to the drug; dose-dependent injection-site reactions occurred in up to 12%.

What the evidence does not show

It does not show that lowering Lp(a) prevents anything. As of the Lp(a)HORIZON design paper in the American Heart Journal, no drug was approved to lower Lp(a) and no randomised trial had shown that lowering Lp(a) reduces cardiovascular risk. That trial is the test: 8323 patients with established cardiovascular disease and Lp(a) of 70 mg/dL or more, randomised to monthly pelacarsen or placebo, with major adverse cardiovascular events as the primary endpoint, a minimum of 2.5 years of follow-up, and completion pegged to the accrual of 993 confirmed events - an anticipated total duration of about six years. Percentage reductions in a biomarker are a promissory note. Events are the payment, and the payment has not cleared.

Two further absences are worth naming because the internet fills them in anyway. The EAS consensus statement, having reviewed the evidence, does not support Lp(a) as a risk factor for venous thrombotic events. And while high Lp(a) is associated with aortic valve micro- and macrocalcification and with incident stenosis, nothing in this evidence base tests whether lowering Lp(a) changes a valve's trajectory.

What to order, and what a result changes now

• Ask for it by name. It is not on a routine lipid panel. NHLBI notes that providers may order it for people with a family history of early heart disease or stroke.
• Check your units before you read your result. The NLA classifies under 75 nmol/L (30 mg/dL) as low risk, 75 to 125 nmol/L (30 to 50 mg/dL) as intermediate, and 125 nmol/L (50 mg/dL) or higher as high risk. The two scales are not interchangeable, and the same sample can look alarming on one and unremarkable on the other if you read it against the wrong ruler.
• Once, not annually. "At least once" is what both bodies recommend. In the absence of an approved drug that moves the number, there is no guideline case for serial monitoring.
• Then spend the information on the modifiable factors. The EAS follow-up paper addresses the "why measure it" objection head-on: its risk calculator shows Lp(a) influences lifetime risk of atherosclerotic cardiovascular disease, and that global risk may be substantially underestimated in people with high or very high Lp(a). The recommended response is intensifying management of the other risk factors, LDL-C first - which is a conversation with a clinician, not a self-service decision.

The reason this test is interesting right now is that it is one of the few places in consumer longevity testing where the science is ahead of the shelf rather than behind it. Lp(a) has causal genetics, an agreed threshold structure, more than a fifth of the world's cardiovascular disease population carrying an elevated level, and three drugs that can flatten it almost completely. What it does not have is a single randomised outcome. If you get tested, get tested knowing that: you are buying a lifetime risk estimate that sharpens everything else on your chart, and a place in the queue for a class of drugs that has not yet proved it works.