The First Lp(a)-Lowering Drug Failed Its Big Trial: What That Means If Your Lp(a) Is High

September 6, 2026

On Friday afternoon Novartis announced that pelacarsen, the first drug built to lower lipoprotein(a), did not prevent heart attacks, strokes, or cardiovascular deaths in the trial that was supposed to settle the question. It lowered Lp(a) the way it was designed to. It failed anyway. People who took it had the same rate of events as people who took a placebo.

In July I wrote a preview of these drugs and said that if the trials came back disappointing, it would be a hard but valuable lesson, and far better to learn it from a trial than from treating patients on a hunch. I did not expect to be writing the follow-up this soon, and I did not expect this result. Most of the field didn’t. Novartis stock dropped more than 7 percent shortly after the announcement, and Amgen’s fell almost as much on a drug that wasn’t the one being tested.

If your Lp(a) is high, or someone in your family has a high number, you probably have questions. I’ll go through what was announced, what may have gone wrong, what this does and does not change about the risk itself, and what I’d tell you to do this week.

What the trial found

Lp(a)HORIZON was the trial’s name, and it was large. It enrolled 8,323 people at centers around the world, all of whom had already had a heart attack, a stroke, or symptomatic narrowing of the leg arteries, and all of whom had an Lp(a) of at least 70 mg/dL, which works out to roughly 150 nmol/L in the units I wrote about in June. A prespecified group with levels of 90 mg/dL and up, roughly 190 nmol/L, sat above the 175 nmol/L line where pooled NIH data showed the risk persisting in people who were already being treated well. These were the patients the drug was meant for.

Half were given pelacarsen as a monthly injection under the skin, 80 milligrams, and half were given a placebo injection. Everyone stayed on standard treatment for their disease, statins and the rest, and the trial was built to run until 993 people had suffered one of the events being counted. It was planned to take about six years, with everyone followed for at least two and a half. An event meant death from a cardiovascular cause, a non-fatal heart attack, a non-fatal stroke, or an urgent hospital procedure to open a blocked heart artery. Pelacarsen brought Lp(a) down. Ionis, the company that invented the drug, said the lowering was consistent with earlier studies, where this dose cut Lp(a) by roughly 70 to 80 percent. And it did not reduce the number of events in the overall study group. Novartis has not released the numbers. That is all we know. We do not yet know how many events occurred in each arm, whether there was a small difference that fell short of statistical proof, or what happened in the prespecified group whose Lp(a) was 90 mg/dL or higher. All of that comes when the full results are presented at a medical meeting, which has not been named.

Novartis’ chief medical officer said these were “not the results we hoped for.” Ionis’ chief executive said the company was disappointed that the lowering “did not translate to cardiovascular risk reduction.” Both statements went out after the stock market closed on a Friday, the traditional slot for news a company would rather not discuss.

Why the whole field expected this to work

Lp(a) is a cholesterol-carrying particle your liver makes according to instructions in a single gene. You inherit your level, it reaches its adult value in early childhood and stays there, and diet, exercise, and statins leave it about where it was. Roughly one in five people has a high level, most without knowing it. People with high Lp(a) have more heart attacks, more strokes, and more narrowing of the aortic valve, and the higher the number the worse the odds. I cover all of that in my guide to lipoprotein(a). The case that Lp(a) actually causes this damage, as opposed to traveling alongside it, rests on genetics. People born with gene variants that give them high Lp(a) for life have more heart disease. People born with variants that give them low Lp(a) have less. Nature ran the experiment for us, across millions of people, and the answer came back the same way every time. That kind of evidence is about as strong as it gets short of a trial, and it is the reason drug companies spent billions building medicines to turn Lp(a) production down.

There was one more piece of encouragement. PCSK9 inhibitor injections, the ones used to lower LDL, happen to lower Lp(a) by a quarter or so as a side effect, and when researchers looked back at those trials, the patients whose Lp(a) fell the most seemed to do a little better than their LDL alone would predict. In one of them, people who started with higher Lp(a) got about three times the event reduction of people who started lower. That fell well short of proof. It pointed the right way.

So the logic was clean. A causal risk factor, a drug that removes most of it, and a trial to count the events.

Friday’s lesson is that clean logic and a positive trial are two different things.

Where this leaves the theory

Let me say the plain thing first. We were surprised.

I was, and so were most of the cardiologists whose reactions I read over the weekend. If you had polled my field a month ago, the large majority would have bet on this trial working, and the debate would have been about how big the benefit was, not whether there was one. When a result lands that far from what the experts expected, the right response is to take the theory apart and look at each piece.

The full data are not out, so everything below is a hypothesis. I am going to walk through them in order, from the reading I find most depressing to the one I actually believe.

The most depressing possibility is that Lp(a) is only a marker

In the darkest reading, Lp(a) is a surrogate for something else. The genetics tell us that people who inherit high Lp(a) have more heart attacks. They do not tell us, by themselves, that the Lp(a) particle is the thing doing the damage. The same stretch of DNA could be doing something else we have not measured, or the particle could be a bystander that reliably travels with the true culprit. If that is the case, our theories about how Lp(a) causes disease, that it carries cholesterol into plaque, inflames the artery wall, and may interfere with the body’s system for dissolving clots, are at best fundamentally incomplete, and at worst completely wrong.

Cardiology has been here before, more than once.

Niacin lowered Lp(a) by 20 to 40 percent, and raised HDL, and did nothing for events when it was added to a statin, in trials that were never built around Lp(a) in the first place. Drugs built to raise HDL, the so-called good cholesterol, raised it dramatically and mostly prevented nothing, and the one that showed a small benefit did so by lowering LDL. Lowering homocysteine with B vitamins, a sure thing in the 1990s, went the same way, with no fewer heart attacks or deaths. Each of those started with a marker that tracked with disease, a theory of why, and a drug that moved the marker. Each ended with a trial that said the theory was wrong.

I do not think this is the most likely reading. The genetic case for Lp(a) is stronger and more specific than the case for HDL or homocysteine ever was, and the particle itself turns up inside plaque and inside narrowed aortic valves. A single negative trial cannot settle it either way. It is the possibility that should keep all of us humble, and it is the one I would be arguing with myself about if the next two trials come back negative too.

The knockdown threshold, a more hopeful possibility

A more hopeful reading is that Lp(a) is so harmful that a 70 to 80 percent reduction still leaves enough of it to keep causing events at a rate the trial could not distinguish from placebo. Call it the knockdown threshold hypothesis.

Genetic studies have tried to estimate how much you would need to lower Lp(a) to get the same benefit as a modest LDL reduction, and the estimates are sobering. Depending on the method, they land between about 65 and about 100 mg/dL in absolute terms, which is a bigger drop than most people in this trial could have achieved, since you cannot remove more Lp(a) than you started with. Someone entering at 75 mg/dL who loses 75 percent has lost 56. That may not be enough. There is a catch buried in those estimates, and I will come back to it, which is that they describe a lifetime spent at the lower level, not a few years.

If this is the answer, the two drugs behind pelacarsen have a real shot, because they cut deeper. Amgen’s olpasiran lowers Lp(a) by more than 95 percent and Lilly’s lepodisiran by about 94 percent, against the 70 to 80 percent pelacarsen delivers. Analysts landed on this explanation within hours, and it happens to be the explanation that keeps the other drugs in play, which is a reason to hold it loosely. It is testable, and it will be tested.

There is a smaller cousin of this idea worth a sentence. Everyone in the trial was on modern therapy, and Novartis pushed the expected results back by about a year in early 2025 because events were accruing more slowly than planned, which is the signature of a well-managed group. When the background rate of events is low, a drug has to be very good to show a difference on top of it, and a real but modest benefit can hide inside the statistical noise. Event counts will tell us whether there was a trend the trial was too small to confirm.

My favorite theory is that Novartis went the wrong way

My own reading is that the trial asked the question backwards. Novartis went for what looked like an easy layup, a secondary prevention indication, meaning people who already had heart disease. It is the standard playbook, and it makes business sense. People who have already had a heart attack have events at a high rate, so you need fewer patients and fewer years to show a difference, and if it works you have a drug for the sickest, most obviously deserving group. The catch is that these are the people whose arteries are already full of plaque. Decades of it.

The way I think about LDL explains why that worries me. I have argued on this site that what damages your arteries is cumulative exposure to LDL, the total amount your bloodstream has carried over a lifetime, the area under the curve, and that the payoff from lowering it is largest when you start young, decades before the plaque has formed. I suspect the same logic applies to Lp(a), and applies with more force. Your LDL drifts up over adulthood. Your Lp(a) reaches its adult level in early childhood. A 62-year-old in this trial with an Lp(a) of 90 mg/dL has carried that level since he was a small boy, and the disease that put him in the trial is the finished product of that exposure. Taking the Lp(a) away in year sixty-three may be too late to change what six decades already built. This is the catch in the genetic estimates I mentioned above. They measure what a lifetime at a lower level does, which is the one thing a trial that starts at 62 cannot deliver.

If that is right, the real utility of these drugs is in primary prevention. The patients most likely to benefit may be the youngest ones with the highest levels and no events at all, and the patients in this trial, with serious vascular disease and a heart full of plaque, may have been the ones least able to benefit. The exposure has not finished its work in a 40-year-old, and stopping it could change the trajectory the way starting a statin at 35 changes it. That is the trial I most want to see, and nobody is running it. Two trials now include people who have not had an event, one from Amgen and one from Lilly, and both are a step in the right direction. Both start at age 50 or 55, though, in people who already carry other risk factors or visible plaque. Nobody has yet enrolled the 35-year-old.

What to watch next

Four trials will decide this, and they answer different questions.

Start with the full Lp(a)HORIZON data. Novartis will present the complete results at a medical meeting it has not yet named. I will be looking at how far apart the two groups were, what happened in the prespecified group with an Lp(a) of 90 mg/dL or higher, whether the people whose Lp(a) fell the furthest did any better than the rest, and whether the curves started to separate late, which would hint that time, not the drug, was the missing ingredient. A trial that failed by a whisker means something different from one that failed flat.

Next is OCEAN(a)-Outcomes, Amgen’s olpasiran trial, which is the direct test of the knockdown threshold. Olpasiran cuts Lp(a) by more than 95 percent with an injection every three months, and the trial enrolled 7,297 people who have already had a heart attack or a stent and whose Lp(a) was 200 nmol/L or higher, a higher bar than the Novartis trial used. It is again a secondary prevention trial, so it does not test my favorite theory. It is fully enrolled and listed to finish in 2028, and Amgen has not said when it will report. If deep knockdown in the sickest patients works where 70 to 80 percent did not, the threshold idea wins.

Then ACCLAIM-Lp(a), Lilly’s lepodisiran trial. Lepodisiran is given every six months and lowers Lp(a) by about 94 percent. This is the one I am watching most closely, because it enrolls people who have never had an event, alongside people who have, more than 17,300 in all, with an entry threshold of 175 nmol/L, the same line I wrote about in June. The people without prior events had to be at least 55 and carry other risk factors. If they benefit and the people with prior events do not, that pattern would say more about Lp(a) than any single number could, and it would point every future trial toward younger patients. It is listed to finish in 2029.

Amgen is asking the same question. OCEAN(a)-PreEvent started enrolling in 2025 and is built entirely around people who have not had a heart attack or stroke, about 11,000 of them, age 50 and up, with Lp(a) of 200 nmol/L or higher and other risk factors or visible plaque. It is the cleanest test of the primary-prevention idea anyone has designed, and it is listed to finish in 2031. That is a long wait.

Behind those four, Lilly’s daily pill, muvalaplin, has an outcomes trial of its own underway in more than 10,000 people, and a smaller company, Silence Therapeutics, has a fourth injection that has not yet reached that stage.

If most of these trials come back positive, Friday will be remembered as the wrong drug, the wrong dose, or the wrong patients. If they come back negative, we will have learned something about Lp(a) that the genetics could not tell us, and I will write that post too.

What this does not change

The risk itself is where I expect the most confusion. This trial tested a drug. It did not test whether Lp(a) is dangerous. The genetic evidence that high Lp(a) raises your risk of heart attack, stroke, and aortic valve disease is exactly as strong today as it was on Thursday.

Your number did not change. What changed is that the first attempt to treat it directly did not work.

That means the testing advice stands. The 2026 American cholesterol guideline recommends that every adult have Lp(a) measured once in their life, and the European guideline says the same. I wrote about that guideline in June and I would say the same thing today. If you have never had the test, ask for it with your next blood draw. If your level is high, your parents, siblings, and children should be tested too, because about half of them will share it. One test, once, and you know something about your risk that no other test would have told you.

And it means the plan for a high number stands. Everything I laid out in June for patients above 175 nmol/L still applies, and I have not changed it. A lower LDL target than your risk score alone would give you, because LDL is the part of the picture we can move and lifetime exposure to it tracks with the risk of dying from heart disease. Blood pressure treated to goal. No tobacco. A coronary calcium score if you have not had one, to see how much disease has already built up. Friday’s news leaves that indirect strategy as the only one we have. Do it well.

If you are in one of the other trials

If you are enrolled in one of the other Lp(a) outcomes trials, please do not stop on the strength of this news. Amgen’s drug and Lilly’s drug are different molecules, given on different schedules, lowering Lp(a) further, in different groups of patients, and two of the trials are asking the primary-prevention question this one never touched. Friday’s result answered one question, and those trials are asking others. Your study team will be talking with you about this in the coming weeks, and that conversation is the right place for your questions. Stopping a study drug early, on your own, is the one way to make sure your years in the trial answer nothing.

All of those trials are continuing, as of this writing.

My take

The strange position I described in July, measuring a number that nothing treats, is unchanged. What changed is that we now have one answer to the question of whether treating it would help, and the answer was no, at least for this drug, this dose, and these patients. I am less certain than I was in July that lowering Lp(a) will turn out to prevent heart attacks. I am no less certain that high Lp(a) causes them. Those are different claims, and the gap between them is where the next two years of research will live.

My bet, for what it is worth, is that the trial tested the right idea in the wrong people. Lp(a) is a lifetime exposure, and the place to interrupt a lifetime exposure is early. If the Lilly trial shows a signal in the people who had not yet had an event, that bet pays off. Should the Amgen trial show instead that deeper knockdown works in the sickest patients, the threshold idea pays off, and I will happily be wrong about which one. And if neither shows anything, the depressing reading moves to the front of the line.

So test once. Treat everything around the number as hard as your risk warrants. Hold the drugs to the standard of a positive trial, which is the standard that just caught the first one. If you want to talk through your own number, whether that is with me here in Encinitas or with your own cardiologist, bring the actual lab value, in whichever units your lab reported, and we can go from there. The nmol/L number is the more useful one if you have both.

Frequently asked questions

Did the pelacarsen trial fail?

Yes. On September 4, 2026, Novartis announced that pelacarsen did not meet the primary endpoint of the Lp(a)HORIZON trial. It lowered Lp(a), but people who took it had no fewer cardiovascular deaths, heart attacks, strokes, or urgent artery-opening procedures than people who took a placebo. Detailed numbers have not been released and will be presented at a medical meeting.

Does this mean high Lp(a) is not dangerous after all?

No. This trial tested a drug, not the risk factor. Genetic evidence that high Lp(a) raises the risk of heart attack, stroke, and aortic valve narrowing is unchanged. What the trial showed is that lowering Lp(a) with this drug, in these patients, did not reduce events. Why that happened is not yet known.

Why would lowering Lp(a) fail to prevent heart attacks?

Nobody knows yet. The most worrying possibility is that Lp(a) is a marker for some other cause we have not identified, which would mean the theory behind these drugs is incomplete or wrong. Two more hopeful possibilities are that a 70 to 80 percent reduction is not deep enough and the newer drugs that cut Lp(a) by 95 percent will succeed, or that the trial treated people whose arteries were already damaged by a lifetime of exposure and the benefit lies in treating younger people before their first event. Full data and the trials still running should narrow the list.

Could an Lp(a) drug still work if it is started earlier, before a heart attack?

That is the theory I find most convincing. Lp(a) reaches its adult level in early childhood, so a patient who has a heart attack at 60 has carried a high level for six decades, and the plaque that caused the event is the finished product of that exposure. Lowering Lp(a) late may not undo it. Lowering it in a 35-year-old with a very high level and no events could change the whole trajectory, the way starting a statin early does for LDL. Two trials, Lilly’s ACCLAIM-Lp(a) and Amgen’s OCEAN(a)-PreEvent, include people without prior events, though both start at age 50 or 55, and their results are not expected before 2029 and 2031.

Should I still get my Lp(a) tested?

Yes. The 2026 American cholesterol guideline and the current European guideline both recommend that every adult have Lp(a) measured once in their lifetime. A high result changes how aggressively your other risk factors should be treated and flags relatives who should be tested. The result of this trial does not change that advice.

What should I do if my Lp(a) is high?

The plan is the same one that applied before the trial. Lower your LDL cholesterol further than your risk score alone would call for, treat blood pressure to goal, do not use tobacco, and consider a coronary calcium score to measure how much plaque is already present. Have your first-degree relatives tested. There is still no approved drug that targets Lp(a) itself.

I am enrolled in one of the other Lp(a) trials. Should I drop out?

No. The Amgen and Lilly trials are testing different drugs that lower Lp(a) more deeply, on different dosing schedules, in different patient groups, two of them in people who have never had an event, and all of them are continuing. Talk with your study team about the news. Stopping a study drug on your own makes your participation unable to answer the question the trial was built to answer.

Will any Lp(a) drug ever be approved?

That depends on the trials still running. Amgen’s olpasiran is being tested in people with prior heart attacks and very high Lp(a), with more than 95 percent lowering, in a trial listed to finish in 2028, and separately in people without prior events, in a trial listed to finish in 2031. Lilly’s lepodisiran is being tested in a group that includes both, listed to finish in 2029. If any of those succeeds, an approved Lp(a) drug is still possible before the end of the decade. If they all fail, the idea that lowering Lp(a) prevents events will be in real doubt.

References

1. Novartis. “Novartis Announces Lp(a)HORIZON Phase III Topline Results for Pelacarsen in Patients with Elevated Lp(a) and Established Cardiovascular Disease (CVD).” Media release, Basel, September 4, 2026.

2. Ionis Pharmaceuticals. “Ionis Partner Novartis Announces Lp(a)HORIZON Phase 3 Topline Results for Pelacarsen in Patients with Elevated Lp(a) and Established Cardiovascular Disease (CVD).” News release, September 4, 2026.

3. Cho, Leslie, Stephen J. Nicholls, Børge G. Nordestgaard, Ulf Landmesser, Sotirios Tsimikas, Michael J. Blaha, et al. “Design and Rationale of Lp(a)HORIZON Trial: Assessing the Effect of Lipoprotein(a) Lowering with Pelacarsen on Major Cardiovascular Events in Patients with CVD and Elevated Lp(a).” American Heart Journal 287 (2025): 1-9.

4. Tsimikas, Sotirios, Ewa Karwatowska-Prokopczuk, Ioanna Gouni-Berthold, et al. “Lipoprotein(a) Reduction in Persons with Cardiovascular Disease.” New England Journal of Medicine 382, no. 3 (2020): 244-255.

5. Parhofer, Klaus G., Ulrich Julius, Anna Laura Herzog, et al. “Pelacarsen and Lipoprotein(a) Apheresis in Secondary Prevention: The Lp(a)FRONTIERS APHERESIS Trial.” European Heart Journal 47, no. 25 (2026): 3284-3294.

6. Kinzelmann, Fabienne, and Madison Muller. “Novartis Drug Fails Key Test for New Heart Disease Approach.” Bloomberg, September 4, 2026.

7. Fidler, Ben. “Novartis, Ionis Drug Failure Spurs Questions About an Emerging Class of Heart Medicines.” BioPharma Dive, September 5, 2026.

8. Reyes-Soffer, Gissette, Henry N. Ginsberg, Lars Berglund, et al. “Lipoprotein(a): A Genetically Determined, Causal, and Prevalent Risk Factor for Atherosclerotic Cardiovascular Disease: A Scientific Statement from the American Heart Association.” Arteriosclerosis, Thrombosis, and Vascular Biology 42, no. 1 (2022): e48-e60.

9. Burgess, Stephen, Brian A. Ference, James R. Staley, et al. “Association of LPA Variants with Risk of Coronary Disease and the Implications for Lipoprotein(a)-Lowering Therapies: A Mendelian Randomization Analysis.” JAMA Cardiology 3, no. 7 (2018): 619-627.

10. Lamina, Claudia, and Florian Kronenberg, for the Lp(a)-GWAS-Consortium. “Estimation of the Required Lipoprotein(a)-Lowering Therapeutic Effect Size for Reduction in Coronary Heart Disease Outcomes: A Mendelian Randomization Analysis.” JAMA Cardiology 4, no. 6 (2019): 575-579.

11. O’Donoghue, Michelle L., Sergio Fazio, Robert P. Giugliano, et al. “Lipoprotein(a), PCSK9 Inhibition, and Cardiovascular Risk.” Circulation 139, no. 12 (2019): 1483-1492.

12. O’Donoghue, Michelle L., Robert S. Rosenson, Baris Gencer, et al. “Small Interfering RNA to Reduce Lipoprotein(a) in Cardiovascular Disease.” New England Journal of Medicine 387, no. 20 (2022): 1855-1864.

13. Nissen, Steven E., Wei Ni, Xi Shen, et al. “Lepodisiran - A Long-Duration Small Interfering RNA Targeting Lipoprotein(a).” New England Journal of Medicine 392, no. 17 (2025): 1673-1683.

14. Nicholls, Stephen J., Wei Ni, Grace M. Rhodes, et al. “Oral Muvalaplin for Lowering of Lipoprotein(a): A Randomized Clinical Trial.” JAMA 333, no. 3 (2025): 222-231.

15. Blumenthal, Roger S., Pamela B. Morris, Mario Gaudino, et al. “2026 ACC/AHA/AACVPR/ABC/ACPM/ADA/AGS/APhA/ASPC/NLA/PCNA Guideline on the Management of Dyslipidemia.” Circulation (2026). doi:10.1161/CIR.0000000000001423.

16. Mach, François, Konstantinos C. Koskinas, Jeanine E. Roeters van Lennep, et al. “2025 Focused Update of the 2019 ESC/EAS Guidelines for the Management of Dyslipidaemias.” European Heart Journal 46, no. 42 (2025): 4359-4378.

17. Amgen. “Olpasiran Trials of Cardiovascular Events and Lipoprotein(a) Reduction (OCEAN(a)) - Outcomes Trial.” ClinicalTrials.gov NCT05581303.

18. Amgen. “OCEAN(a)-PreEvent - Olpasiran Trials of Cardiovascular Events and Lipoprotein(a) Reduction to Prevent First Major Cardiovascular Events.” ClinicalTrials.gov NCT07136012.

19. Eli Lilly and Company. “A Study to Investigate the Effect of Lepodisiran on the Reduction of Major Adverse Cardiovascular Events in Adults with Elevated Lipoprotein(a) (ACCLAIM-Lp(a)).” ClinicalTrials.gov NCT06292013.

20. Family Heart Foundation. “Ongoing Clinical Trials Targeting Lipoprotein(a).” Updated March 20, 2026.

Published on damianrasch.com. The above information was composed by Dr. Damian Rasch, drawing on individual insight and bolstered by digital research and writing assistance. The information is for educational purposes only and does not constitute medical advice.