Lp(a)HORIZON: What Pelacarsen’s Failure Really Teaches Us About Lipoprotein(a)
On September 4, 2026, Novartis announced the topline results of Lp(a)HORIZON, the largest cardiovascular outcomes trial ever run on lipoprotein(a), or Lp(a). Pelacarsen, an antisense drug built to switch off Lp(a) production in the liver, lowered Lp(a) levels as expected. But it did not cut the rate of heart attacks, strokes, cardiovascular death, or urgent revascularization compared with placebo. The trial missed its primary endpoint.
If you are a GP, you have probably heard about Lp(a) testing over the past few years, maybe from a cardiology referral letter, a patient who read about it online, or a professional update. This result matters to you directly, because it changes what you should tell patients about testing, and what you should expect from treatment in the near term. I want to walk you through what happened, why it happened, and what you should actually do differently in your clinic starting now.
What the trial did
Lp(a)HORIZON enrolled 8,323 patients across the world. Every patient had established cardiovascular disease and an Lp(a) level of at least 70 mg/dL, a level found in roughly one in five people globally. Patients received either 80 mg of pelacarsen by injection once a month or a placebo injection, on top of their existing statin, ezetimibe, or PCSK9 inhibitor therapy. The study also built in a planned subgroup analysis of patients with even higher Lp(a), at least 90 mg/dL, to check whether a higher starting point would show a clearer benefit.
The drug worked exactly as designed on the biomarker. Pelacarsen is a GalNAc-conjugated antisense oligonucleotide, which is a technical way of saying it is a small strand of genetic material that gets delivered straight to liver cells and blocks the instructions for building apolipoprotein(a), the protein that makes Lp(a) unique among lipoproteins. In the earlier phase II study, weekly doses cut Lp(a) by up to 80%, and in Lp(a)HORIZON, the monthly 80 mg dose again produced a large and sustained reduction. Novartis has not yet released the exact percentage from the phase III trial, since full data will come at a medical congress later this year, but the company has confirmed the drug lowered Lp(a) substantially and had an acceptable safety profile, with injection site reactions as the main side effect.
Despite that biological success, the four-point composite endpoint of cardiovascular death, non-fatal heart attack, non-fatal stroke, and urgent hospitalized revascularization showed no significant difference between pelacarsen and placebo, in the overall group and, as far as has been disclosed, in the higher-risk 90 mg/dL subgroup as well.
Why this is a genuine surprise, not a shrug
For over a decade, the case for treating Lp(a) rested on strong genetic and observational evidence. Mendelian randomization studies, which use naturally occurring genetic variation to test cause and effect, consistently showed that people born with lifelong high Lp(a) have a higher rate of coronary disease and aortic valve narrowing. That kind of evidence usually predicts that lowering the same molecule with a drug will help. It worked that way for LDL cholesterol. Lp(a)HORIZON was the trial meant to confirm it works the same way for Lp(a).
It is worth being honest here: this result does not prove Lp(a) is irrelevant to cardiovascular risk. It proves that lowering Lp(a) with this drug, in this population, for this length of follow-up, did not translate into fewer events. There are a few real explanations worth holding in mind rather than one simple story.
First, the background care in this trial was excellent. Average LDL cholesterol in the study population sat around 65 to 66 mg/dL, and most patients were on high-intensity statins with many on a second lipid-lowering drug. When the rest of a patient’s cardiovascular risk is already controlled that tightly, there is less room left for any single additional intervention to show a measurable benefit, simply because the pool of preventable events shrinks.
Second, follow-up in cardiovascular outcomes trials for a slow-building disease like atherosclerosis may need more than the roughly 2.5 to 6 years used here. Lp(a)’s damage builds over decades, and a few years of suppression in patients who already have established disease might come too late to reverse existing plaque burden, even if it can eventually be shown to help with earlier prevention.
Third, this may be a case where the biomarker and the disease process are not as tightly linked in the treated state as the genetic data suggested. Genetic studies capture a lifetime of exposure, from birth onward. A drug started in a 60-year-old with known heart disease is not the same experiment as being born with low Lp(a).
What this means for your clinic today
Do not change your testing advice. Lp(a) testing once in a lifetime for every adult remains recommended by both American and European guidelines, because Lp(a) is still one of the strongest independent, genetically fixed risk factors for coronary disease we know about, and knowing a patient’s level helps you have an honest conversation about their overall risk and how aggressively to manage everything else you can control: LDL cholesterol, blood pressure, smoking, and diabetes.
Do not offer pelacarsen, because it is not approved and, based on this result, is unlikely to seek approval on the strength of this trial alone. If a patient asks about it after reading a headline, tell them plainly that the drug lowers the number on the blood test but has not yet been shown to lower their actual risk of heart attack or stroke, and that this is genuinely useful information rather than bad news, because it stops patients chasing an unproven treatment.
Keep doing what already works. For a patient with high Lp(a), your job has not changed. Push LDL cholesterol down as low as guidelines recommend, control blood pressure, treat diabetes properly, and support smoking cessation. These interventions have trial evidence behind them, and elevated Lp(a) is one more reason to be strict about the risk factors you can actually modify today.
Manage expectations, not just numbers. A patient with a high Lp(a) result, especially one who found out through a relative’s genetic history, often wants a specific plan aimed at that one number. It is reasonable and kind to explain that we do not yet have a proven drug for that specific number, while reassuring them that overall risk reduction still protects them.
The pipeline has not stopped
Lp(a)HORIZON is a setback, not the end of the Lp(a) story. Several other drugs using different chemistry and different dosing schedules are still being tested in ongoing outcomes trials, including lepodisiran and zerlasiran, which are small interfering RNA, or siRNA, drugs that work through a related but distinct mechanism, olpasiran, and an oral small-molecule option called muvalaplin. These trials run longer follow-up in some cases and enroll somewhat different populations, so it is possible one of them will show a benefit that pelacarsen did not, particularly if timing of treatment relative to disease progression turns out to matter as much as I suspect it does. It is also possible the whole approach needs rethinking, and that will only become clear once more of these outcomes trials read out over the next two to three years.
My honest take
I think this result should make the field more careful, not more cynical. A biomarker can be a real and independent risk factor and still fail to produce a treatable target on the first attempt, especially in a population already receiving near-optimal care. That happened before with HDL cholesterol raising drugs, which looked promising on paper for years before repeated outcomes trials showed no benefit. Lp(a) may or may not follow the same path. What I would resist is either extreme: dismissing Lp(a) as irrelevant, or continuing to treat it as an established therapeutic target when the actual outcomes evidence so far says otherwise.
For now, your best move as a GP is to keep testing Lp(a) once per patient as guidelines recommend, use a high result as motivation to be strict about every other risk factor you can control, and wait for the full congress presentation and the results from the other ongoing Lp(a) trials before changing anything else in your practice. This is not a small update. It is one of the more important negative trials in preventive cardiology in years, and it deserves to be read carefully rather than reduced to a single headline.
Sources: Novartis media release, September 4, 2026; Ionis Pharmaceuticals investor release, September 4, 2026; ClinicalTrials.gov NCT04023552; coverage from HCPLive, TCTMD, and Neurology Advisor, September 2026.
