The short version
On September 4, Novartis announced that pelacarsen — the first drug ever designed to lower Lp(a), an inherited cardiovascular risk factor — did not reduce heart attacks, strokes, or cardiovascular deaths in its big clinical trial.
The drug did lower Lp(a) itself. It just didn’t translate that lowering into fewer events.
The trial, called Lp(a)HORIZON, enrolled 8,323 people who had already had a heart attack, stroke, or serious peripheral artery disease, and who had high Lp(a).
The trial’s design had real limits. It didn’t measure inflammation. It didn’t image patients’ plaque. It didn’t distinguish between the two very different kinds of Lp(a) particles people can carry. And it may not have run long enough for a lifetime disease to respond.
Two other Lp(a) drugs, from different companies, are still being studied. Their trials are ongoing.
For patients: nothing about your care changes today. High Lp(a) is still a risk factor worth knowing. It still means we treat everything else — cholesterol, blood pressure, lifestyle — more aggressively.
For the science: the Lp(a) story is bruised, not buried.
On September 4, 2026, Novartis announced that pelacarsen, the first drug ever engineered to lower a substance called lipoprotein(a), did not reduce heart attacks, strokes, cardiovascular deaths, or urgent heart procedures in its big Lp(a)HORIZON clinical trial. The drug did what it was designed to do at the chemistry level. It lowered Lp(a) substantially, just as it had in every earlier study. But that lowering did not translate into fewer cardiovascular events over the years the trial ran.
A lot of well-meaning coverage this week has already collapsed the story into “Lp(a) doesn’t matter” or “the Lp(a) hypothesis is dead.” Neither is true. I want to walk you through what HORIZON actually tested, what it didn’t test, and why I think the more interesting part of the story is still ahead of us.
What Lp(a) is, in one paragraph
Lp(a) — pronounced “L P little a” — is a cholesterol-carrying particle in your bloodstream. Everyone makes some. How much you make is set almost entirely by your genes, which is why Lp(a) runs in families and why diet and exercise barely move it. People with high Lp(a) have roughly two to four times the lifetime risk of heart attack and stroke that people with low Lp(a) have. For decades, we had no way to lower it. Pelacarsen was the first drug built specifically to try.
What the HORIZON trial did
HORIZON enrolled 8,323 patients who had already had one of three serious events: a heart attack, an imaging-confirmed stroke, or symptomatic peripheral artery disease (blockages in the leg arteries). Everyone also had a high Lp(a) — at least 70 mg/dL, with a typical enrolled value of about 108 mg/dL.
Each patient was randomly assigned to get either a monthly injection of pelacarsen or a placebo injection. Both groups also stayed on all their usual heart medications — statins, blood pressure pills, and so on. The average patient in the trial already had a well-controlled LDL cholesterol of about 65 mg/dL. Patients were followed for anywhere from four to about six and a half years, depending on when they enrolled.
The trial’s main question was straightforward: would people who got the drug have fewer heart attacks, strokes, cardiovascular deaths, or urgent heart procedures than people who got the placebo? The answer, in the topline data released last week, was no. The two groups came out about the same.
What HORIZON did not measure
This is the part I want you to sit with, because it shapes how much weight the negative result should carry.
HORIZON measured Lp(a) itself, and it collected the usual safety labs — blood counts, kidney function, liver enzymes. That was essentially it on the biology side. It did not measure inflammation markers like hsCRP. It did not measure Lp-PLA2 or oxidized LDL, both of which reflect the inflammatory damage happening in blood vessels. It did not measure the oxidized fatty acids that Lp(a) is thought to carry into the artery wall — even though an earlier, smaller pelacarsen study showed reductions in those markers of 37 to 88 percent.
And HORIZON did not image anyone’s arteries. No serial CT scans. No ultrasounds of the coronaries. No PET scans of vascular inflammation. Nothing to tell us whether pelacarsen actually changed the plaque in patients’ arteries.
Think of it this way. We ran a five-year trial on whether a new drug prevents heart attacks. It didn’t. But we never checked whether the drug did anything to the plaques that cause heart attacks, or to the inflammation that drives them. All we know is that the number on the lab report went down and the clinical events didn’t.
That’s a real gap. Compare it to how the cholesterol drug evolocumab was studied. In the GLAGOV trial, researchers put tiny ultrasound probes inside patients’ coronary arteries before and after treatment. They could actually see the plaque shrinking. In the HUYGENS trial, they used a different high-resolution imaging technique and saw the plaques becoming more stable and less inflamed. By the time evolocumab’s big outcomes trial reported that it prevented events, we already had a chain of evidence linking the drug to the artery wall. HORIZON gives us no such chain.
The units problem, and the isoforms problem
There’s one more thing HORIZON didn’t do that deserves its own section — because it may mean the wrong patients were enrolled.
Lp(a) can be reported in two different units, and those two units measure two different things.
Milligrams per deciliter (mg/dL) measures the weight of Lp(a) particles in your blood.
Nanomoles per liter (nmol/L) measures the number of Lp(a) particles in your blood.
Weight and number are not the same thing. Imagine measuring a crowd of people by their total weight versus by counting heads. If some of them are children and some are adults, weight and headcount give you very different answers. The 2022 European Atherosclerosis Society consensus and the 2024 National Lipid Association scientific statement both recommend measuring Lp(a) by particle number, in nmol/L. Particle number is what actually gets stuck in the artery wall. Weight is a rough substitute that can mislead you.
HORIZON enrolled patients based on weight — mg/dL — measured at a central lab. When the design paper mentions nmol/L values, those are just arithmetic conversions, not separate measurements.
Now to the second problem. Lp(a) particles are not all the same size. They have a tail called apo(a) that varies from person to person. Some people inherit small tails. Some inherit large ones. This isn’t just a cosmetic difference:
People with small tails carry more particles for the same weight — meaning their true Lp(a) count is higher than the weight number suggests.
People with small tails also seem to have Lp(a) that is more damaging to arteries, particle for particle. A large meta-analysis of 40 studies and 58,000 people found that people with small apo(a) tails had about twice the risk of coronary heart disease compared with people who had large tails.
HORIZON did not measure tail size. It did not sort patients by tail size. It did not analyze small-tail patients separately from large-tail patients. That means the trial mixed together two biologically different groups under a single weight-based cutoff.
Here’s what that could mean in practice. Picture two patients, both enrolled at exactly 90 mg/dL of Lp(a). The one with small tails may have been carrying a much higher number of much more dangerous particles. The one with large tails may have been carrying fewer, less dangerous particles. Both got the same drug at the same dose. Both counted equally in the analysis. If the drug’s benefit depends on how much of the dangerous biology a patient actually has — and there’s every reason to think it does — the small-tail patients may have shown a real benefit that got washed out by the large-tail patients until the average looked like nothing.
This is not a theoretical worry. It is a real reason a negative HORIZON does not close the question of whether pelacarsen works in the patients whose biology it was designed to address.
Why time may still change the answer
Cardiovascular trials are, in a real sense, races between the biology of the drug and the calendar of the trial.
Statins showed their benefit quickly — within a year or two — because they act on both cholesterol and vessel inflammation at the same time. Some other lipid drugs have needed longer.
Lp(a) builds up in artery walls over decades. The people in HORIZON had already accumulated a lifetime of plaque before their qualifying heart attack or stroke. Asking a monthly injection to reverse or even neutralize that burden in four to six years, on top of well-controlled cholesterol, is a demanding test. It’s entirely possible that a five-year curve looked flat while a ten-year curve would separate. It’s also possible that the trial needed younger patients, or patients earlier in their disease, or patients whose cholesterol wasn’t already so aggressively controlled that there was little event risk left to reduce.
Sometimes clinical trial curves overlap for years and then quietly begin to fan apart at the very end. When Novartis presents the full data at the American Heart Association meeting in November, three things will tell us how alive the Lp(a) hypothesis really is.
Did the “harder” version of the primary endpoint — heart attack, stroke, and cardiovascular death, without the softer “urgent procedure” count — move in favor of the drug?
Did the sicker patients, with even higher Lp(a) levels above 90 mg/dL, do better on the drug?
What was the on-treatment Lp(a) level, and how many patients got it down low enough to matter?
What the next generation of trials should do differently
If I could rewrite the Lp(a) research agenda from where we stand today, I would ask four things of the trials that follow.
Build in mechanistic endpoints. Take pictures of the plaques with serial CT or PET imaging in a subset of patients. Measure inflammation and oxidation markers before, during, and after treatment. A negative outcomes trial paired with positive plaque and inflammation data would tell a very different story from a negative outcomes trial with no biology at all.
Enroll on the right measurement. Use particle number (nmol/L), not weight (mg/dL). Sort patients by tail size. Analyze small-tail patients separately from large-tail patients. A trial built on the wrong measurement may weaken its own signal before the first patient is dosed.
Test the hypothesis earlier in the disease. HORIZON studied patients whose arteries were already badly damaged. Two other Lp(a) drugs — olpasiran from Amgen and lepodisiran from Lilly — are still being tested. Both use a newer technology that lowers Lp(a) more deeply and more durably than pelacarsen does. But those trials also enroll patients who have already had events. The truly definitive test would be a trial in high-Lp(a) patients earlier in life, before the first heart attack.
Take follow-up seriously. Event-driven trials that stop after five years are efficient. They are not always long enough for a disease that took fifty years to build.
Where this leaves us in the office
For patients I care for with a high Lp(a), nothing about the HORIZON result changes what I do this month.
Elevated Lp(a) is still an independent, inherited, causal cardiovascular risk factor. Knowing your Lp(a) still matters, because it changes how aggressively we treat everything else — cholesterol, blood pressure, inflammation, the lifestyle levers. A high Lp(a) still means the LDL target should be lower, the blood pressure target tighter, the calcium score more closely watched.
What has changed is our timeline for having a targeted therapy — a drug I can point to and say, “This lowers your Lp(a) and lowers your risk.” That answer is turning out to be more complicated than a single trial can settle. The Lp(a) story is bruised. It is not buried. And the next two trials, with better drugs, longer follow-up, and I hope more careful measurements built into the design, will tell us far more than HORIZON did about what to do with a number we can now measure but still cannot fully treat.
The honest posture, for now, is patience. We have the first real answer. And we have every reason to keep asking the question.
This article is educational and reflects the author’s clinical perspective. It is not medical advice, and it is not a substitute for evaluation and treatment by your own physician. If you have questions about your Lp(a) level or cardiovascular risk, please discuss them with your doctor.
For women-specific cardiovascular content, see the sister publication The Menopause Heart.