The Biggest Heart Drug Bet in a Decade Just Went Bust
Novartis's pelacarsen crushed Lp(a) levels by 80% in over 8,000 patients, but heart attacks and strokes kept happening anyway. The landmark Lp(a)HORIZON trial failure threatens not just one drug, but the entire genetic thesis behind a multi-billion-dollar drug class.
The Science Said It Should Work
For decades, cardiologists had a villain they couldn't catch. Lipoprotein(a), or Lp(a), is a cholesterol-like particle that circulates in your blood and clogs your arteries. Unlike regular cholesterol, you can't lower it with statins or diet. Your Lp(a) level is almost entirely determined by your genes, and roughly one in five people are walking around with dangerously high levels.
The genetics were compelling. Study after study showed that people born with high Lp(a) had more heart attacks, more strokes, more of everything bad. Mendelian randomization studies (a technique that uses genetic data to mimic a clinical trial) found that every 10 mg/dL increase in genetically determined Lp(a) raised coronary heart disease risk by about 5.8%. People with two-fold higher genetic Lp(a) levels had roughly 22% more heart attacks. The dose-response curve was clean. The signal was loud.
The logic felt bulletproof: if high Lp(a) causes heart disease, then lowering Lp(a) should prevent it. Novartis and Ionis Pharmaceuticals built a $1 billion-plus bet on that logic. Their drug, pelacarsen, was designed to silence the gene responsible for Lp(a) production.
Then 8,323 patients later, the answer came back: it didn't work.
What Just Happened
Novartis announced that pelacarsen failed the primary endpoint of its massive Phase 3 trial, called Lp(a)HORIZON. This was the biggest cardiovascular outcomes trial of its kind, designed to answer a question that cardiology has debated for over 30 years.
The trial enrolled patients between 18 and 80 years old who already had established cardiovascular disease (prior heart attacks, strokes, or peripheral artery disease) and elevated Lp(a) levels of at least 70 mg/dL. On average, participants were about 60 years old, and nearly three-quarters were male. They received either pelacarsen (80 mg monthly injection) or a placebo.
The primary endpoint was a composite of the scariest things that can happen to a heart patient: requiring hospitalization. Think of it as a scorecard for "did fewer terrible cardiac events happen?"
Pelacarsen did exactly what it was supposed to do at the molecular level. It crushed Lp(a) levels, lowering them by up to roughly 80% in earlier Phase 2 data. But all that Lp(a) reduction translated into... nothing. No meaningful reduction in heart attacks. No fewer strokes. No lives saved versus placebo.
It's like fixing a smoke detector that kept going off, only to discover the house was still on fire.
Why This Rocks the Entire Field
This isn't just a bad day for Novartis. It's an earthquake for every company chasing Lp(a).
The whole Lp(a)-lowering drug class was built on the same genetic foundation that pelacarsen just failed to validate clinically. If lowering Lp(a) by 80% doesn't reduce cardiac events, what does that mean for the next wave of drugs that lower it by 90% or more?
Amgen's olpasiran, a different type of drug called an siRNA (small interfering RNA), has shown Lp(a) reductions exceeding 90% in early trials. Its own massive outcomes trial, OCEAN(a)-Outcomes, is ongoing. Eli Lilly has two horses in the race: lepodisiran, another siRNA that showed an 88.5% Lp(a) reduction at one year after a single dose, and muvalaplin, the first oral Lp(a)-lowering pill, which is still in earlier development.
These drugs are mechanistically different from pelacarsen. While pelacarsen is an antisense oligonucleotide (ASO) that recruits one enzyme (RNase H1) to chop up the Lp(a) gene's messenger RNA, siRNAs like olpasiran and lepodisiran use a different intracellular machine called RISC to do the same job. They tend to lower Lp(a) more deeply and for longer.
That distinction matters. Proponents will argue that maybe pelacarsen just didn't lower Lp(a) enough. An 80% reduction sounds dramatic, but some Mendelian randomization analyses estimated that you'd need Lp(a) to fall by roughly 90 mg/dL (about 200 nmol/L) to get the same event reduction as a meaningful statin benefit. If baseline levels were very high, even an 80% cut might not cross that therapeutic threshold.
The "Maybe It's Not Dead" Argument
There's a version of this story where the Lp(a) hypothesis survives. It goes something like this:
Pelacarsen lowered Lp(a) a lot, but the siRNAs lower it a lot more. If olpasiran can push reductions above 95%, maybe that final 15% of suppression is what separates noise from signal. Biology is nonlinear; sometimes you need to cross a threshold before benefits appear.
There's also the question of timing. Lp(a) may contribute to atherosclerosis (artery plaque buildup) over a lifetime. Mendelian randomization studies capture the effect of lifelong low Lp(a). Treating a 60-year-old patient for a few years might not undo decades of arterial damage. It's the difference between never letting weeds grow versus trying to pull them out of a garden that's been neglected for 40 years.
Plus, the trial design had some noise baked into it. Event accrual was slower than expected (patients weren't having heart attacks as fast as anticipated), which is why the readout was pushed back. Slower events can dilute a drug's apparent effect, especially in a trial that's already walking a statistical tightrope.
The Market Fallout
Investors had been bracing for volatility around this readout, and a miss is squarely bad news for Novartis in the near term. Pelacarsen was supposed to be a major cardiovascular growth driver. Without positive outcomes data, the drug has no clear path to blockbuster status.
But the ripple effects extend well beyond Basel. Amgen, Eli Lilly, and every biotech with Lp(a) ambitions now face a credibility problem. Payers and prescribers were already going to need convincing that a new injectable for a biomarker most doctors don't even test for was worth the cost. Now they'll need even more convincing.
The pressure on Amgen's OCEAN(a)-Outcomes trial just doubled. If olpasiran also fails to reduce events, the Lp(a) hypothesis is effectively dead as a therapeutic strategy (at least for secondary prevention in established heart disease). If it succeeds, the narrative flips to "pelacarsen was the wrong drug, not the wrong target."
Either way, the next few years will be defining ones for cardiovascular medicine.
The Bottom Line
Lp(a) had everything going for it: robust genetics, clean dose-response data, a plausible mechanism involving plaque formation, inflammation, and blood clotting. It was the poster child for genetically validated drug targets. And it still failed in a clinical trial.
That's the humbling reality of drug development. Genetics can tell you where to aim. Only outcomes trials can tell you if you've hit the target. Novartis aimed carefully, shot straight, and still missed. The question now is whether someone else can find a bigger gun, or whether the target was never really there at all.
Clinical & Regulatory6 min read
A Blood Test Just Changed How We Treat Breast Cancer
The FDA just approved a breast cancer drug with a twist: treatment decisions are driven by a blood test, not a scan. AstraZeneca's camizestrant could rewrite the rules of precision oncology, but it almost didn't survive a brutal advisory committee vote.
