Lowering Lipoprotein(a): ApoB, PCSK9 Inhibitors and New Drugs in 2026

On 4 September 2026, Novartis announced the result of a trial that much of cardiology had been waiting for. A total of 8,323 people with established cardiovascular disease and markedly elevated lipoprotein(a) had received either pelacarsen or placebo. Pelacarsen did what it was designed to do: it lowered Lp(a). Yet the trial missed its actual goal. Taken together, cardiovascular deaths, heart attacks, strokes and urgent procedures on the coronary arteries did not become measurably less frequent.

The result struck a nerve because the genetic data on Lp(a) look so clear: people born with high levels have more heart attacks and more narrowed aortic valves. A drug lowers the level, yet a reduction in events has not been demonstrated statistically? That does not mean Lp(a) is harmless, nor that every new Lp(a) drug will fail. But it shows very precisely how wide the gap between a lower lab value and a prevented heart attack can be.

This article follows that question from the biology of the particle through ApoB and established cholesterol-lowering drugs to the new Lp(a) agents. The editorial cut-off is 3 November 2026. Up to that date, Novartis had released only a short topline statement: no effect sizes, no safety data, no detailed analyses. This text does not replace medical advice and contains no treatment or dosing recommendation. Novartis, 2026, NLA, 2026

The short answer

• Lp(a) is largely inherited. More than 90 percent of the differences between people are explained by a single gene. European and American societies recommend measuring it at least once in adulthood; from about 50 mg/dL it counts as a risk enhancer.
• Diet and exercise barely lower Lp(a). They still matter, because risk depends on all factors together: LDL cholesterol or ApoB, blood pressure, smoking, diabetes.
• In the EU, no drug is approved specifically to lower Lp(a) (as of 3 November 2026). PCSK9 inhibitors lower LDL cholesterol by roughly half to almost two thirds and Lp(a) moderately, by a little over a quarter in the FOURIER trial. Their benefit on heart attacks is proven but cannot be attributed to Lp(a) lowering alone.
• New agents lower Lp(a) by 70 to almost 100 percent in phase 2 trials. The first large outcome trial, Lp(a)HORIZON with pelacarsen, missed its main goal according to the topline statement. Trials of olpasiran, lepodisiran and muvalaplin are registered to run until at least 2028.

A particle with an extra arm: what lipoprotein(a) is

Cholesterol is fat-soluble, so it cannot simply float in watery blood. The body packs it into tiny transport particles called lipoproteins. The best known is LDL, low-density lipoprotein. An LDL particle has a core of cholesterol compounds and fats, a shell of phospholipids, and a single large protein wrapped around it like a belt: apolipoprotein B-100, or ApoB-100.

Lipoprotein(a), pronounced “L-P little a”, is essentially such an LDL particle. But it carries an extra protein, apolipoprotein(a), firmly attached to ApoB-100 by a disulfide bridge, a chemical link between two sulfur atoms. Apo(a) is built from folded units called kringles, named after a Scandinavian pastry. One kringle type, kringle IV type 2, is repeated several times in a row, and the number of repeats is inherited. Kronenberg et al., 2022

• ApoB-100: exactly one molecule per particle
• Particle surface with lipid core
• Disulfide bridge between ApoB-100 and apo(a)

Circles on apo(a): kringle units. The number of kringle IV type 2 repeats differs by inheritance; short apo(a) forms usually go with much higher Lp(a) levels. Schematic, not to scale, not a molecular structure.

Why genes decide almost everything

More than 90 percent of Lp(a) levels are determined by variants in the LPA gene. The number of kringle IV type 2 repeats alone explains roughly 30 to 70 percent of the variation, depending on the population. Short apo(a) forms with few repeats lead to much higher levels. The range is accordingly wide: from below 0.1 to above 300 mg/dL. The European consensus estimates that around 1.4 billion people worldwide have elevated Lp(a). Kronenberg et al., 2022

The level is not entirely fixed. Women have on average about 5 to 10 percent higher levels, and these tend to rise after menopause. Kidney disease can raise Lp(a) considerably, three- to fivefold in nephrotic syndrome. These are shifts around an inherited baseline, though. Someone born with very high Lp(a) almost always keeps a high level.

What Lp(a) does to arteries and the heart valve

Why does this matter? Like LDL, Lp(a) carries cholesterol into the artery wall. It also transports oxidised phospholipids, altered fat molecules that can trigger inflammation in the vessel wall. Because apo(a) resembles the clotting-system protein plasminogen, Lp(a) is also suspected of slowing the breakdown of blood clots; so far the evidence for this comes mainly from laboratory studies. According to the European consensus, Lp(a) is not a risk factor for venous thrombosis. Kronenberg et al., 2022

The strongest evidence comes from genetics. An analysis in UK Biobank compared genetic variants that specifically raise Lp(a) with variants that raise LDL and scaled both to the number of ApoB-carrying particles. Particle for particle, Lp(a) was about six times as strongly linked to coronary heart disease as LDL (point estimate 6.6; 95% confidence interval 5.1 to 8.8). Björnson et al., 2024

The aortic valve carries a genetic fingerprint too. A genome-wide study with computed tomography data from 6,942 people found a variant in the LPA gene that roughly doubled the odds of a calcified aortic valve per copy. In the Swedish cohort, the same variant was associated with a 68 percent higher risk of incident aortic stenosis per copy; the association was replicated in a Danish cohort. Thanassoulis et al., 2013

Because such gene variants are allocated at random at conception, they strongly suggest that Lp(a) itself contributes to disease rather than merely accompanying it. The European Atherosclerosis Society consensus describes the genetic evidence for a causal link with atherosclerosis and aortic stenosis as extensive. Keep one thing in mind for later: this evidence describes the consequences of a lifelong higher or lower level, not the consequences of treatment that starts at age 60.

Why LDL cholesterol does not tell the whole story

Most lab reports show LDL cholesterol, or LDL-C. It indicates how much cholesterol is carried in LDL particles: a mass, not a particle count. It is a little like weighing the cargo of all delivery vans on a road instead of counting the vans. Usually the two match well. But if the particles are small and carry little cholesterol, more vans are on the road for the same cargo.

ApoB as a measure of particle number

This is where ApoB comes in. Every lipoprotein that can drive atherosclerosis carries exactly one ApoB molecule: VLDL, IDL, LDL and Lp(a). The laboratory measures ApoB mass concentration rather than counting individual particles. Because each of these particles carries one ApoB molecule, this serves as a measure of their total number. In 2024, an expert consensus of the National Lipid Association concluded that ApoB reflects risk better than LDL-C, both before and during cholesterol-lowering treatment. Despite a strong correlation across the population, the two values can diverge in individuals. When such discordance occurs, risk tends to follow ApoB or non-HDL cholesterol. Soffer et al., 2024

The 2026 American guideline uses ApoB selectively: to assess residual risk in people with cardiovascular-kidney-metabolic syndrome, type 2 diabetes, high triglycerides or known cardiovascular disease who have already reached their LDL-C and non-HDL goals. It does not position ApoB as a routine replacement for LDL-C. ACC, 2026

Where Lp(a) fits in

Lp(a) particles carry ApoB and are therefore included in the ApoB count. As a rule, though, they make up only a small share of all ApoB particles. The ApoB level therefore says little about whether the many particles include a large number of the genetically unfavourable Lp(a) particles. Conversely, an Lp(a) result says nothing about the other particles. The two measures complement each other; neither replaces the other.

• LDL cholesterol

  Measures: the amount of cholesterol in LDL particles.

  Reveals: the target of almost all guidelines; the basis of most treatment trials.

  Does not reveal: how many particles are circulating, or whether many of them are Lp(a).

• ApoB

  Measures: ApoB mass concentration as a proxy for particle number (one ApoB per particle, Lp(a) included).

  Reveals: residual risk when LDL-C and particle number diverge, for example in diabetes or high triglycerides.

  Does not reveal: what share of the particles is Lp(a).

• Lp(a)

  Measures: the concentration of Lp(a) particles, ideally in nmol/L.

  Reveals: a largely inherited, independent risk factor; one measurement is usually enough.

  Does not reveal: overall risk, which depends on all factors together.

This leads to a point that often gets lost in consultations: a good LDL-C result does not rule out Lp(a)-related risk. And one more detail for interpreting lab results: what is reported as LDL cholesterol usually also includes the cholesterol carried in Lp(a) particles. With very high Lp(a), LDL-C therefore does not solely represent cholesterol in LDL particles themselves. Nevertheless, the consensus advises against routine mathematical correction: the cholesterol content of Lp(a) varies substantially. Kronenberg et al., 2022

When and how Lp(a) is measured

The recommendations now largely agree. The European Atherosclerosis Society consensus recommends measuring Lp(a) at least once in adults, plus targeted family testing, for example in familial hypercholesterolaemia or when relatives have had early heart attacks or high Lp(a). The 2026 American guideline likewise recommends measuring it at least once in adulthood; because lifestyle barely changes the level, repeat testing is usually unnecessary. Kronenberg et al., 2022, ACC, 2026

The 2025 update of the ESC/EAS guideline puts it this way: Lp(a) above 50 mg/dL (about 105 nmol/L) should be considered a risk-enhancing factor in all adults, with higher levels associated with greater risk (class IIa, level of evidence B). The American guideline names 50 mg/dL or more as a risk enhancer; in its assessment, 50 mg/dL or more goes with about 1.4 times, and 100 mg/dL or more with about twice, the risk of atherosclerotic cardiovascular disease. ESC/EAS, 2025, AHA, 2026, ACC/AHA Guideline, 2026

mg/dL or nmol/L: why conversion is tricky

Lp(a) is reported in two units. mg/dL gives the mass of the whole particle, nmol/L the number of particles. Because apo(a) varies in size and weight depending on its number of kringles, an Lp(a) particle does not weigh the same in everyone. There is therefore no fixed conversion factor that holds for all. The documents themselves show it: the 2025 ESC/EAS guideline equates 50 mg/dL with about 105 nmol/L, while the 2022 EAS consensus gives 50 mg/dL and 125 nmol/L as parallel risk thresholds, not an individual conversion formula. Societies prefer nmol/L and assays that are as independent of apo(a) size as possible. When comparing two results, check the unit and the laboratory rather than converting yourself.

What a high result means, and what it does not

A high Lp(a) is not a diagnosis and not a verdict. It is a risk enhancer that adds to the rest of the picture, and risk rises continuously with the level rather than jumping at a threshold. At very high levels above 180 mg/dL (above 430 nmol/L), the additional lifetime risk is, according to the EAS consensus, about equivalent to that of untreated heterozygous familial hypercholesterolaemia. What a high result means for an individual depends on everything else: age, LDL-C, blood pressure, smoking, diabetes, kidney function, family history. That overall assessment belongs with the treating physician. Kronenberg et al., 2022

What can actually be done today

Lifestyle: little effect on Lp(a), large effect on overall risk

The honest answer first: diet and exercise hardly change Lp(a). The European consensus summarises the data in a table. Replacing saturated fat with carbohydrates or unsaturated fats actually raises Lp(a) by about 10 to 15 percent, while a low-carbohydrate diet high in saturated fat lowers it by about 15 percent. Physical activity and fasting have practically no effect. Kronenberg et al., 2022

That sounds paradoxical, but it is not. Swapping out saturated fat lowers LDL cholesterol, and LDL involves far more particles than Lp(a). A slightly higher Lp(a) with a clearly lower LDL-C is not a bad trade for overall risk. With inherited high Lp(a), the factors that can be changed matter all the more: not smoking, blood pressure and blood sugar on target, a healthy body weight and regular exercise. How to progress training sensibly and sustainably is covered in our article on progressive overload.

Lowering LDL: statins, ezetimibe, bempedoic acid

For high Lp(a), the societies so far mainly recommend one thing: treat the remaining risk early and consistently and, where appropriate, lower LDL cholesterol more intensively, depending on absolute risk and Lp(a) level. Whether targeted lowering of Lp(a) itself prevents heart attacks has not yet been shown, according to the EAS. EAS, Lipoprotein(a)

Statins are the foundation because they have reduced cardiovascular events in many large trials. Their effect on Lp(a) is not favourable, though how large it is remains disputed. A meta-analysis of individual data from 5,256 participants in six trials found Lp(a) about 11 percent higher on statins than on placebo. An analysis of seven large statin outcome trials cited by the EAS, by contrast, found no effect. In practice this changes little: the EAS explicitly states that people with high Lp(a) and sufficiently high risk should take statins. Tsimikas et al., 2020, EAS, Lipoprotein(a)

Ezetimibe reduces cholesterol absorption from the gut and is usually combined with a statin. Bempedoic acid inhibits an enzyme of cholesterol synthesis in the liver, earlier in the pathway than statins. In the CLEAR Outcomes trial of 13,970 people who could not or would not take statins because of side effects, it lowered LDL-C after six months by 21 percentage points more than placebo. Over a median of 40.6 months, the composite endpoint occurred in 11.7 versus 13.3 percent (hazard ratio 0.87). Gout (3.1 versus 2.1 percent) and gallstones (2.2 versus 1.2 percent) were more frequent. The 2025 ESC/EAS guideline names ezetimibe, PCSK9 antibodies and bempedoic acid as non-statin therapies with proven cardiovascular benefit. Nissen et al., 2023, ESC/EAS, 2025

Lipoprotein apheresis: blood cleaning for a few

The only method that currently removes Lp(a) from the blood directly and strongly is lipoprotein apheresis. Blood is filtered outside the body, Lp(a) and LDL are removed, and the treatment is repeated regularly. The EAS consensus sees it as an option for very high Lp(a) with progressive cardiovascular disease; large randomised trials demonstrating prevention of heart attacks or deaths are lacking.

In Germany, access is tightly regulated. Under the criteria described, for example, by the Association of Statutory Health Insurance Physicians North Rhine, apheresis for isolated Lp(a) above 60 mg/dL with LDL-C in the normal range is only considered if progressive cardiovascular disease is documented clinically and by imaging. An advisory commission reviews every application, and approval is usually limited to one year. For the vast majority of people with elevated Lp(a), apheresis is therefore not an option. KV Nordrhein, 2026

A small randomised crossover trial in 20 people with refractory angina assessed apheresis using myocardial perfusion and symptoms. It does not replace large trials of heart attacks or mortality. Khan et al., 2017

PCSK9 inhibitors: the detour via the LDL receptor

The liver pulls LDL out of the blood with its own grabbing arms, the LDL receptors. A receptor binds an LDL particle, draws it into the cell and then returns to the surface to catch the next one. The protein PCSK9 interrupts this cycle: it attaches to the receptor and ensures that it is degraded along with its cargo instead of returning. Fewer receptors mean more LDL in the blood. Blocking PCSK9 extends the life of the receptors, and the liver clears more LDL.

Antibodies and siRNA: two routes to the same goal

There are two principles. The monoclonal antibodies evolocumab and alirocumab capture PCSK9 in the blood; they are injected under the skin, evolocumab according to the EMA about every two weeks or once a month. Inclisiran, by contrast, is a small interfering RNA, or siRNA: it enters liver cells and causes the blueprint for PCSK9 to be broken down before the protein is made. After the first injection, a second follows after three months, then one every six months. EMA, Repatha, EMA, Leqvio

Both lower LDL-C strongly. In the FOURIER trial of 27,564 people with atherosclerotic disease, evolocumab lowered LDL-C by 59 percent, from a median of 92 to 30 mg/dL. In the registration trials ORION-10 and ORION-11, with almost 3,200 participants together, inclisiran lowered LDL-C by about 50 percent versus placebo after 510 days. ApoB falls along with LDL-C, because fewer particles remain in the blood. Sabatine et al., 2017, Ray et al., 2020

Where outcome data exist and where they do not

For both antibodies, clinical benefit has been shown in large trials. In FOURIER, the composite endpoint occurred after a median of 2.2 years in 9.8 percent on evolocumab and 11.3 percent on placebo (hazard ratio 0.85). In ODYSSEY OUTCOMES, with 18,924 people after an acute coronary syndrome, the figures were 9.5 versus 11.1 percent on alirocumab (hazard ratio 0.85). In 2025, the VESALIUS-CV trial showed benefit for the first time in 12,257 people with atherosclerosis or diabetes but without a previous heart attack or stroke: estimated five-year rates of a first major event were 6.2 percent on evolocumab and 8.0 percent on placebo; median follow-up was 4.6 years (hazard ratio 0.75). Sabatine et al., 2017, Schwartz et al., 2018, Bohula et al., 2026

For inclisiran, such proof is still missing. The European Medicines Agency notes that there is no direct evidence yet of fewer heart attacks or strokes; the outcome trials ORION-4 and VICTORION-2 PREVENT are ongoing. EU approvals differ accordingly: evolocumab and alirocumab are also approved to reduce cardiovascular risk in established atherosclerotic disease, inclisiran for treating raised blood lipids. EMA, Leqvio, Mafham et al., 2026

Common side effects include injection-site reactions: 3.8 versus 2.1 percent in ODYSSEY, and 2.6 to 4.7 percent versus under one percent in the ORION trials. VESALIUS-CV showed no difference from placebo in safety events.

Funding matters: FOURIER and VESALIUS-CV were funded by Amgen, and ODYSSEY OUTCOMES by Sanofi and Regeneron. These large randomised trials provide important clinical evidence; manufacturer funding still belongs in its interpretation. Sabatine et al., 2017, Schwartz et al., 2018, Bohula et al., 2026

What PCSK9 inhibitors do to Lp(a), and what that proves

This is where it gets interesting for our topic. PCSK9 inhibitors also lower Lp(a), but moderately. In FOURIER, the median reduction after 48 weeks was 26.9 percent. In ODYSSEY OUTCOMES, Lp(a) fell by a median of 5 mg/dL from a median baseline of just over 21 mg/dL. Why a drug that acts on the LDL receptor lowers Lp(a) is not fully understood. O’Donoghue et al., 2019, Bittner et al., 2020

The secondary analyses are intriguing. In FOURIER, people with Lp(a) above the median benefited more: their risk fell by 23 percent (hazard ratio 0.77), compared with 7 percent at lower Lp(a) (hazard ratio 0.93, not statistically significant). The absolute difference is even clearer: over three years, evolocumab prevented an event in 2.49 of every 100 treated people in the higher-Lp(a) group and in 0.95 of every 100 in the lower-Lp(a) group. Arithmetically, 40 and 105 people respectively had to be treated for three years to prevent one event. The statistical test for a true difference between the groups narrowly missed the usual threshold, however (P = 0.07). In ODYSSEY OUTCOMES, Lp(a) lowering predicted lower risk independently of LDL lowering. O’Donoghue et al., 2019, Bittner et al., 2020

This is an interesting signal, but not proof that Lp(a) lowering produced the benefit. People with high Lp(a) start at higher risk and therefore gain more in absolute terms from any effective risk reduction, including the much larger LDL reduction. And both analyses are secondary analyses within trials that were not randomised on Lp(a). The American guideline still draws a practical conclusion: in established atherosclerotic disease, Lp(a) of 50 mg/dL or more and goals not met on maximally tolerated statin therapy, it specifically recommends adding a PCSK9 antibody with proven cardiovascular benefit. The rationale is the proven LDL lowering and event reduction, not a demonstrated effect of Lp(a) lowering. AHA, 2026, ACC/AHA Guideline, 2026

The new generation: agents that target Lp(a) directly

The new agents no longer target clearance but production. Lp(a) is made in the liver, and almost all new approaches interrupt the blueprint for apo(a) there, its messenger RNA. Antisense oligonucleotides are short, synthetic single-stranded nucleic acid chains that bind to this messenger RNA and trigger its breakdown. siRNA agents are double-stranded and use a cutting mechanism the cell already has; their effect often lasts for months. A third approach prevents apo(a) from binding to the LDL particle at all.

The early Lp(a) programmes are also manufacturer-funded: including pelacarsen by Akcea, olpasiran by Amgen, lepodisiran and muvalaplin by Eli Lilly, and zerlasiran by Silence Therapeutics. Short controlled trials can demonstrate large laboratory effects; independent confirmation, long-term safety and clinical outcomes remain separate questions. Tsimikas2020, O’Donoghue2022, Nissen2025, Nissen2024, Nicholls2025

The overview below places the agents side by side. The percentages come from different trials with different participants, time points, assays and calculation methods. They show that all candidates lower Lp(a) strongly, but they are not a head-to-head comparison.

Table sources: Tsimikas et al., 2020, Novartis, 2026, O’Donoghue et al., 2022, OCEAN(a)-Outcomes, Nissen et al., 2025, ACCLAIM-Lp(a), Nissen et al., 2024, Silence Therapeutics, Nicholls et al., 2025, MOVE-Lp(a), CRISPR Therapeutics, 2026

Pelacarsen: the pioneer

Pelacarsen, originally developed by Ionis and licensed to Novartis, was the first Lp(a) agent to enter a large outcome trial. In the phase 2 trial, 286 people with cardiovascular disease and Lp(a) of at least 60 mg/dL (150 nmol/L) received different dosing schedules or placebo for six to twelve months. After six months, Lp(a) was 35 to 80 percent lower depending on the schedule, compared with 6 percent on placebo. Injection-site reactions were the most common side effect. Tsimikas et al., 2020

Lp(a)HORIZON then tested the decisive question. It enrolled people with established cardiovascular disease and Lp(a) of at least 70 mg/dL; a subgroup with at least 90 mg/dL was analysed separately. All received guideline-directed lipid-lowering and blood pressure treatment. The primary endpoint was a bundle of four events: cardiovascular death, non-fatal heart attack, non-fatal stroke and urgent coronary revascularisation requiring hospitalisation. That goal was not reached: the bundle was not measurably reduced. Novartis’s head of development put it this way: the lower Lp(a) did not translate into reduced cardiovascular risk in the overall population. Novartis, 2026

What had not been published by 3 November 2026 is at least as important: how far Lp(a) fell in the trial, the hazard ratio with its confidence interval, absolute event rates, the result in the subgroup at 90 mg/dL or above, secondary endpoints, length of follow-up and safety data. Novartis said the data would be presented at an upcoming medical congress. In its assessment, the National Lipid Association stressed that Lp(a) remains an important, independent risk factor and should still be measured. NLA, 2026

A different benefit was tested in the Novartis-funded Lp(a)FRONTIERS APHERESIS trial published in 2026: among 51 people already receiving regular apheresis, pelacarsen substantially reduced the proportion of sessions needed over 52 weeks compared with placebo. This endpoint does not establish prevention of heart attacks or deaths. Injection-site redness was more frequent. Parhofer et al., 2026

Olpasiran: near-complete lowering

Amgen’s olpasiran is an siRNA. In the phase 2 trial OCEAN(a)-DOSE, 281 people with atherosclerotic disease and Lp(a) above 150 nmol/L received different doses every 12 or 24 weeks. Median baseline Lp(a) was 260 nmol/L. After 36 weeks, Lp(a) was 70.5 to 101.1 percent lower than with placebo depending on dose. Values above 100 percent arise from the statistical comparison with placebo; they do not mean a negative concentration and do not by themselves establish that levels fell below the assay detection limit. Adverse events overall were about as frequent as with placebo; the most common was pain at the injection site. O’Donoghue et al., 2022

The outcome trial OCEAN(a)-Outcomes enrolled 7,297 people with Lp(a) of at least 200 nmol/L who have had a heart attack and/or a catheter procedure on the coronary arteries. It measures coronary deaths, heart attacks and urgent revascularisations. According to the registry as of 27 October 2026, it is no longer recruiting, and completion of the primary analysis is estimated for March 2028. OCEAN(a)-Outcomes

Lepodisiran: one injection, months of effect

Eli Lilly’s lepodisiran first stood out for its duration. In the phase 1 trial with 48 adults without cardiovascular disease, Lp(a) was still a median 94 percent below baseline 337 days after a single 608 mg dose. In the phase 2 trial ALPACA, with 320 participants and Lp(a) of at least 175 nmol/L, 400 mg lowered time-averaged Lp(a) between days 60 and 180 by 93.9 percent versus placebo. Serious adverse events were not attributed to the drug; injection-site reactions occurred in a dose-dependent manner, in up to 12 percent of the highest-dose group. Nissen et al., 2023, Nissen et al., 2025

A secondary analysis presented at the 2026 ESC congress also showed a dose-dependent reduction in the oxidised phospholipids carried by Lp(a). The authors themselves see this as biological rationale, not evidence of clinical benefit. That answer is expected from ACCLAIM-Lp(a): about 17,300 planned participants with established disease or high risk of a first event, with completion estimated in the registry for March 2029. ACC, 2026, ACCLAIM-Lp(a)

Zerlasiran: strong lab effects, outcome evidence pending

Silence Therapeutics’ zerlasiran is also an siRNA. In a phase 2 trial with 178 people with stable atherosclerotic disease and Lp(a) of 125 nmol/L or more, time-averaged Lp(a) to week 36 fell by 81 to 86 percent versus placebo; serious adverse events were not attributed to the drug. In its 2025 annual report, the company described completed phase 3 readiness activities and a search for a partner. This is a dated development plan, not evidence of clinical outcomes. Nissen et al., 2024, Silence Therapeutics

Muvalaplin: the first tablet

Muvalaplin, also from Eli Lilly, is not a genetic medicine but a small molecule taken by mouth. It prevents apo(a) from binding to the ApoB particle, so no complete Lp(a) is formed. In the phase 2 trial with 233 people, 240 mg daily lowered Lp(a) after twelve weeks by 85.8 percent versus placebo when measured with an assay for the intact particle, or by 68.9 percent with a conventional apo(a) assay. That difference is a lesson in itself: with new mechanisms, the assay can shape the result. ApoB fell by 16 percent. No safety concerns emerged over the twelve weeks. The outcome trial MOVE-Lp(a) is recruiting. Nicholls et al., 2025, MOVE-Lp(a)

Genome editing: once and for all?

Genome editing reaches furthest into the future. With CTX320, CRISPR Therapeutics is testing in a phase 1 trial whether the LPA gene can be permanently switched off in liver cells. In January 2026, the company reported reductions of up to 73 percent during dose finding and is already developing a successor. Peer-reviewed efficacy and safety data were not available by 3 November 2026. A permanent change to the genome also demands a far higher safety bar than an injection whose effect wears off. CRISPR Therapeutics, 2026

The real question: lowering is not the same as protecting

Back to Lp(a)HORIZON. Genetics points to a causal role for Lp(a), a drug lowers the level, and a reduction in events has not been demonstrated statistically. How does that fit together? Several explanations are possible, and without the full data none of them can be confirmed or excluded.

Surrogate markers and clinical benefit

A lab value such as Lp(a) is a surrogate marker: a measure that stands in for what really matters, namely heart attacks, strokes and deaths. A surrogate is only as good as the link between it and the event under that particular treatment. For LDL cholesterol, that link is well established by dozens of trials with statins, ezetimibe, PCSK9 inhibitors and bempedoic acid. For drug-induced Lp(a) lowering, it is still missing. That is exactly why outcome trials are needed, and why phase 2 results of 90 or 100 percent cannot be converted into prevented heart attacks.

Genetics is lifelong, a trial lasts years

The genetic data describe people who have had lower or higher levels since birth. Their arteries were exposed differently for decades. A drug, by contrast, starts in people who often already have plaque, and works for a few years. Whether late lowering can measurably offset existing damage within a trial lasting a few years is an open question. With LDL cholesterol, lowering works relatively quickly; for Lp(a), that has not been shown.

How much lowering is enough?

A 2018 Mendelian randomisation analysis using data from 53 studies and more than 80,000 people with coronary heart disease estimated how large an Lp(a) reduction would need to be to match the effect of lowering LDL-C by 38.67 mg/dL (1 mmol/L): about 101.5 mg/dL (95% confidence interval 71 to 137). Burgess et al., 2018

A simple worked example, not a trial result: someone starting at 100 mg/dL who lowers their level by 80 percent achieves an absolute reduction of 80 mg/dL, less than that estimate. Even an impressive percentage reduction can be too small in absolute terms if the starting level is only moderately raised. How far Lp(a) actually fell in Lp(a)HORIZON is not yet known. The estimate also rests on genetic modelling assumptions and cannot be transferred one-to-one to treatment.

Who was treated, and against what background?

All participants in Lp(a)HORIZON were already receiving guideline-directed lipid and blood pressure treatment. The better the remaining risk is controlled, the fewer events occur and the harder it is to detect an additional effect. The trials also differ considerably: Lp(a)HORIZON enrolled from 70 mg/dL, OCEAN(a)-Outcomes only from 200 nmol/L, ACCLAIM-Lp(a) from 175 nmol/L and including people without a previous event. The endpoints are also composed differently; OCEAN(a)-Outcomes, for example, does not include stroke. The result of one trial cannot simply be transferred to the others.

Relative and absolute risks

Interpretation requires relative and absolute effects, the endpoint, follow-up duration and uncertainty. A hazard ratio compares event rates over time; it is not identical to the ratio of two cumulative risks. The relative risk reduction says by what percentage a risk falls; the absolute one says how many of 100 treated people actually avoid an event. The FOURIER example above shows the difference: a relative reduction of 23 percent corresponded to 2.49 avoided events per 100 people over three years. For Lp(a)HORIZON, neither number is known yet. Whether the result was narrowly neutral or a clear null finding therefore cannot be said.

What a negative and a positive result allow you to conclude

A missed primary endpoint permits a narrow statement: with this drug, at this dose, in this population and over this trial period, the prespecified main goal could not be demonstrated statistically. It does not permit the conclusions that pelacarsen had no effect at all, that Lp(a) is unimportant, or that other agents will fail. Individual subgroups or secondary endpoints reported in future can only be read as hints after a missed main goal, not as proof.

Conversely, a positive result for olpasiran or lepodisiran would show that the agent in question reduces events. It would not by itself prove that this is due to Lp(a) lowering alone, because some agents change other values too; muvalaplin, for instance, lowered ApoB by 16 percent. Science learns in both directions, but it only learns correctly if the question a trial actually answered is not confused with the question one would have liked it to answer.

Conclusion: what is proven, what is promising and what remains open

The trial from the opening has made an uncomfortable truth visible: a drug that lowers a risk marker is not yet a drug that prevents heart attacks. That applies to Lp(a) just as it does to any other lab value.

Proven: Lp(a) is largely inherited, easy to measure once in a lifetime and, according to genetic data, an independent, causal risk factor for atherosclerosis and aortic stenosis. A good LDL-C result does not rule out this risk. Statins, ezetimibe, bempedoic acid and PCSK9 antibodies reduce cardiovascular events, For people with high Lp(a), their relevance depends on individual overall risk and the respective indication.

Promising: Several new agents lower Lp(a) by 70 to almost 100 percent, some with a few injections a year or as a tablet, so far without notable safety signals in trials lasting a few months. In FOURIER, people with higher Lp(a) gained more in absolute terms from evolocumab.

Uncertain: whether targeted Lp(a) lowering prevents heart attacks and strokes, in whom, from what starting level and how early it would need to begin. Also uncertain: what the full Lp(a)HORIZON data will show and what role long-term safety will play.

For people with elevated Lp(a), this means today: the result is important information, not destiny. It argues for taking the remaining risk especially seriously, keeping an eye on LDL cholesterol and, where appropriate, ApoB, addressing blood pressure, blood sugar and smoking, and encouraging relatives to get tested. Which treatment makes sense in an individual case can only be decided in a personal cardiovascular risk assessment with the treating physician.

The open questions are precise enough for the ongoing trials to answer them: does near-complete Lp(a) lowering reduce events when the starting level is very high? Does earlier treatment achieve more than late treatment? And will the full Lp(a)HORIZON data show a signal in the subgroup with the highest levels? Until then, the most effective lever remains the one proven for decades: lowering overall risk. More articles are available in our knowledge section.