Analysts at William Blair had a number in their models: $6 billion in peak annual U.S. sales for pelacarsen, Novartis’s antisense oligonucleotide targeting lipoprotein(a). The science looked clean. Lp(a) is genetically determined, causally implicated in atherosclerosis, and virtually unmoved by statins. Pelacarsen lowered it, dramatically. The lipidology community had spent years arguing that Lp(a) was the residual cardiovascular risk that no one was treating. On September 4, 2026, Novartis announced the Phase 3 Lp(a)HORIZON results: 8,323 patients, a composite endpoint of cardiovascular death, non-fatal myocardial infarction, non-fatal stroke, and urgent coronary revascularization requiring hospitalization, and a drug that did exactly what it was designed to do to the biomarker while failing to move the clinical needle.
The drug worked. The trial failed. Those two facts, sitting in the same sentence, should unsettle every cardiovascular development program currently organized around Lp(a) as a surrogate endpoint.
The Consensus That Built a $6 Billion Forecast
Lp(a) earned its standing as a therapeutic target through an unusually coherent body of evidence. Mendelian randomization studies established genetic causality. Large epidemiological cohorts linked elevated Lp(a) to premature coronary artery disease and aortic stenosis independent of LDL. Cardiovascular societies have issued guidance calling for Lp(a) measurement in cardiovascular risk stratification. And unlike LDL, which responds to statins and PCSK9 inhibitors, Lp(a) has no approved pharmacologic intervention, creating what looked to many researchers like an unmet need with a well-validated molecular address.
Pelacarsen, developed through a collaboration between Novartis and Ionis Pharmaceuticals, attacked that address with precision. The drug is an antisense oligonucleotide, also known as an ASO, that reduces hepatic production of apolipoprotein(a), the distinguishing protein component of Lp(a) particles. Phase 2 results showed robust Lp(a) reductions in some dose groups. The field took that number and extrapolated forward toward clinical benefit, the way it had done with LDL before JUPITER and with HDL before niacin taught everyone a painful lesson about the gap between a good-looking molecule and a body that responds to it the way a model predicts.
The Lp(a)HORIZON trial enrolled its first participants in December 2019. By the time topline data arrived in September 2026, nearly seven years of investment, advocacy, and competitive positioning had accumulated around the premise that lowering Lp(a) would reduce cardiovascular events. That premise is now in serious doubt.
Why Biomarker Victories Keep Losing Clinical Wars
The uncomfortable pattern here has precedent. An analysis published in PMC and drawing on data from 220 cardiovascular trials that used primary surrogate endpoints between 1990 and 2011 found that surrogate-based cardiovascular programs frequently diverge from clinical outcomes at the confirmatory trial stage. Surrogate endpoints earn regulatory and commercial traction because they are measurable earlier, cheaper to power, and mechanistically plausible. The mechanistic plausibility is precisely what makes them seductive, and precisely what makes them dangerous when the mechanism proves insufficient on its own.
The HDL hypothesis is the most bruising example. Niacin raised HDL substantially. CETP inhibitors raised HDL dramatically. Neither approach translated into fewer heart attacks, and at least one CETP inhibitor program was halted after the drug was associated with increased cardiovascular events in a Phase 3 outcomes trial. LDL reduction through PCSK9 inhibition did translate, and the field used that success to rehabilitate its confidence in lipid-based biomarkers as clinical proxies. Pelacarsen suggests that confidence was overcorrected.
Lp(a) is causally linked to cardiovascular disease at the population level. That causal link, established through genetics, does not guarantee that pharmacologically reducing Lp(a) in a patient who already has established cardiovascular disease will reduce their subsequent event risk. The Lp(a)HORIZON population enrolled patients with elevated Lp(a) and established CVD, patients who were already on background therapies including statins and, in many cases, PCSK9 inhibitors. Whether residual Lp(a)-mediated risk in that population is pharmacologically reversible over a trial’s timeframe, against that background of aggressive lipid management, was a question the trial was designed to answer. The answer it returned complicates the entire Lp(a) targeting thesis in secondary prevention.
Ionis Pharmaceuticals co-developed pelacarsen., as is. Each of those programs faces a scientific environment that just had its foundational assumption stress-tested at 8,323 patients, across multiple years, with a composite endpoint that captures the events patients and cardiologists care about most.
What the Trial Architecture Cannot Tell Us
The immediate question from trial design circles will be whether Lp(a)HORIZON was powered correctly, enrolled the right population, or ran long enough to detect a clinically meaningful signal. These are not evasions. They are genuine methodological considerations. If the enrolled population had sufficiently controlled residual LDL risk through background statin and PCSK9 therapy, and if Lp(a)-mediated pathology represents a smaller fraction of absolute event risk in that optimally managed group, the trial could have been set up against a lower event rate than expected, reducing statistical power to detect a real but modest treatment effect.
Novartis has not yet released full trial data; the September 4 announcement carried only topline results. The detailed breakdown of the primary endpoint components, the Lp(a) reductions actually achieved in the trial population, the background therapy distribution, and the hazard ratios across subgroups will determine how much interpretive work the field can do with this outcome. A drug that demonstrated substantial Lp(a) reductions in a Phase 2 cohort does not automatically deliver the same reduction across 8,323 patients over years of dosing, with adherence, dropout, and titration in the real world of a global outcomes trial.
But even allowing for every methodological caveat, the trial failed. That is not a caveat. It is the outcome.
The deeper problem is what the cardiovascular development community has built on top of the Lp(a) hypothesis. The William Blair $6 billion forecast was not eccentric optimism. It reflected a consensus among cardiovascular specialists, lipidologists, and development teams that the causal genetics plus the biomarker reduction plus the unmet need equaled a clinical win waiting to be confirmed. That equation now has a missing term, and the programs that have not yet reached Phase 3 need to decide whether they are solving for the same missing variable or whether their mechanism offers something structurally different.
Olpasiran and lepodisiran lower Lp(a) more aggressively than pelacarsen. Whether deeper biomarker reduction translates into clinical benefit that a more modest reduction did not achieve is a hypothesis the field is entitled to test. It is also a hypothesis that just became significantly harder to fund, design, and defend to a skeptical review committee. The Lp(a)HORIZON result did not close the door on Lp(a) as a target. It changed what must be proven to walk through it, and by whom, and at what cost, and against what timeline.
The first sponsor to bring full Lp(a)HORIZON data before an advisory committee will set the interpretive frame for every competing program still in development. Every trial design decision made in the next twelve months, about population selection, background therapy standardization, event rate assumptions, and duration, gets made in the shadow of 8,323 patients who did not benefit.
References
- JAMA, “Lipoprotein(a)-Lowering Agent Fails to Reduce Cardiovascular Events”
- Novartis, “Novartis Announces Lp(a)HORIZON Phase III Topline Results: Pelacarsen in Patients with Elevated Lp(a) and Established CVD” (September 4, 2026)
- Empirical Health, “Lp(a) Lowering Medications: Trial Outcomes and Clinical Implications”
- National Lipid Association, “Top-Line Results from the Phase 3 Lp(a)HORIZON Trial”
- PMC/PubMed Central, “Two Decades of Cardiovascular Trials With Primary Surrogate Endpoints: 1990–2011”
Moe Alsumidaie, MBA, MSF, is founder and Chief Editor of Vanguard Publications, which publishes Clinical Trial Vanguard, Pharma Vanguard and BullScope, and Head of Research at CliniBiz. He has two decades in clinical trial operations and data science, with earlier roles at Genentech, Abbott Vascular and Stanford University Medical Center, and is a guest lecturer in clinical trial sciences at Rutgers University.
