The Drug That Crushed a Risk Factor But Couldn't Save a Single Heart
Novartis and Ionis' pelacarsen obliterated Lp(a) levels in over 8,000 patients but completely failed to prevent heart attacks and strokes. The result doesn't just sink one drug; it threatens the entire billion-dollar thesis that lowering this genetic risk factor can save lives.
A Billion-Dollar Bet on the Wrong Answer
Imagine you're a detective. You find the murder weapon at the scene. You trace it back to a suspect. Every piece of genetic evidence points to this person. You're certain you've got your killer.
Then you remove the suspect from the equation, and people keep dying anyway.
That's roughly what just happened with pelacarsen, and it's shaking the foundations of cardiovascular drug development.
On September 4, 2026, Novartis and Ionis revealed that their blockbuster-in-waiting failed the Lp(a)HORIZON trial, an 8,323-patient Phase 3 study that was supposed to prove a simple idea: lower a dangerous blood particle called lipoprotein(a), and you'll prevent heart attacks and strokes. The drug did exactly what it was designed to do on a biological level. It crushed Lp(a) levels. But the patients didn't get better. The primary endpoint (a composite of cardiovascular death, heart attacks, strokes, and emergency coronary procedures) was a miss.
Wall Street didn't wait for the details. Both Ionis and Novartis shares dropped sharply in after-hours trading.
The Genetic Case That Seemed Bulletproof
To understand why this failure stings so badly, you need to understand how strong the evidence was for targeting Lp(a).
Lp(a) isn't like cholesterol, which fluctuates with your diet and exercise. It's almost entirely genetic: 70% to 90% of your Lp(a) level is hardwired at birth. You can't jog it away or kale-smoothie it into submission. If you drew the short straw genetically, you're stuck with high levels for life.
Researchers used a clever technique called Mendelian randomization, which is essentially nature's version of a clinical trial. Because Lp(a) levels are assigned at conception by genetic variants, scientists could study whether people born with higher Lp(a) had worse cardiovascular outcomes, free from the confounding mess of lifestyle factors. The answer was a resounding yes: each two-fold increase in genetically determined Lp(a) was linked to about 22% higher risk of heart attack.
Get tomorrow's biotech intelligence before your competitors.
Join thousands of biotech professionals who start their day with our free, daily briefing.
Prospective studies, genome-wide association data, case-control analyses; everything pointed in the same direction. Lp(a) wasn't just a bystander at the crime scene. It looked like the cause.
So the logic seemed airtight: lower Lp(a), save lives. Multiple companies poured billions into this thesis. Pelacarsen, an antisense oligonucleotide (a synthetic strand of genetic material that silences a specific gene), was the first to reach the finish line of a major outcomes trial.
Hitting the Target and Missing the Point
The Lp(a)HORIZON trial enrolled patients aged 18 to 80 who already had cardiovascular disease: prior heart attacks, strokes, or symptomatic peripheral artery disease. All of them had Lp(a) levels at or above 70 mg/dL. They received either pelacarsen (80 mg injected once a month) or placebo, on top of whatever standard therapy they were already taking.
The study was event-driven, meaning it would run until enough cardiovascular events accumulated: 993 of them, to be exact. With a minimum follow-up of 2.5 years and an expected total duration of about six years, this was a massive, expensive, carefully designed effort.
In earlier Phase 2 testing, pelacarsen had shown it could slash Lp(a) by up to 80% compared to placebo. That's not a subtle nudge; that's a demolition job on a biomarker. But in the outcomes trial, that dramatic biological effect simply didn't translate into fewer heart attacks, strokes, or deaths.
It's a bit like discovering that draining a swamp doesn't actually stop the mosquitoes. You did the hard part. The swamp is gone. The mosquitoes don't care.
A Whole Drug Class Feels the Tremor
Pelacarsen wasn't just testing a single molecule. It was testing a hypothesis, and now every company chasing that same hypothesis has to answer uncomfortable questions.
Amgen's olpasiran, a potent siRNA injectable, is the other major contender in this space. It's positioned as a longer-acting, potentially more powerful option than pelacarsen. But "more powerful" only matters if lowering the target actually helps patients; the very thing Lp(a)HORIZON just failed to show.
Eli Lilly's muvalaplin takes a different angle entirely. It's an oral pill (the only one in late-stage development) that blocks Lp(a) assembly rather than suppressing its genetic production. In Phase 2 testing, it showed Lp(a) reductions of up to 85.8% at the highest dose. Lilly is now running its own Phase 3 outcomes trial, and the pressure on that readout just went through the roof.
Analysts quoted by Fierce Biotech noted that there may still be a case for deeper Lp(a) inhibition or for targeting patients with even higher baseline levels. The Lp(a)HORIZON trial tested patients at 70 mg/dL and above, with a pre-specified subgroup analysis at 90 mg/dL and above. But the topline announcement didn't break out those subgroup results in detail, leaving the field to speculate about whether the problem is the drug, the dose, or the entire theory.
Ionis CEO Brett Monia acknowledged the disappointment, saying the result would "meaningfully inform future cardiovascular care." That's the kind of phrase executives reach for when there's no silver lining to polish.
The Surrogate Endpoint Trap
This failure is a painful reminder of one of drug development's oldest lessons: lowering a biomarker is not the same as curing a disease.
The history of cardiology is littered with examples. Drugs that raised "good" cholesterol (HDL) flopped in outcomes trials. Drugs that lowered blood sugar didn't always prevent the complications of diabetes. Biology is messier than a single number on a lab report.
Lp(a) lowering may still matter in certain populations, at certain magnitudes, or through certain mechanisms. But the clean, elegant story (high Lp(a) causes heart disease, lower it and you prevent events) just got a lot more complicated.
What Happens Next
The full data from Lp(a)HORIZON, including hazard ratios, subgroup analyses, and safety details, haven't been released yet. Those details could reshape the narrative. Maybe pelacarsen's monthly dosing wasn't potent enough. Maybe the patient population was too broad. Maybe there's a subgroup where it worked.
But "maybe" is cold comfort when you've spent years and billions on a trial that was supposed to be a coronation. For Ionis, the financial sting is real but contained: the company's 2026 revenue guidance sits at roughly $890 million, buoyed by other pipeline assets like Tryngolza (olezarsen), which is projected to bring in about $105 million in net product sales.
For the broader Lp(a) field, the conversation has shifted overnight. It's no longer "can we lower Lp(a)?" (we obviously can) but rather "does lowering Lp(a) actually save lives, and if so, how much do you need to lower it, in whom, and by what mechanism?"
Those are harder questions. And right now, nobody has the answers.
Clinical & Regulatory4 min read
A Blood Pressure Drug Just Slowed Lung Scarring. Wait, What?
United Therapeutics' Tyvaso, already approved for pulmonary hypertension, just aced two Phase 3 trials in a completely different disease: the deadly lung-scarring condition IPF. The results are reshaping a competitive landscape that patients desperately need shaken up.
