A Heart Drug Slashed a Genetic Risk Factor — but Didn’t Prevent Heart Attacks

A drug that did precisely what it was designed to do to a blood marker failed to help patients’ hearts. Novartis announced on September 4, 2026 that pelacarsen missed the primary endpoint of the Phase 3 Lp(a)HORIZON trial, despite successfully lowering lipoprotein(a).

What Lp(a) is and why it looked like a sure thing

Lipoprotein(a) is an LDL-like particle carrying an additional protein, apolipoprotein(a). Blood levels are overwhelmingly genetically determined — set largely at birth, barely responsive to diet, exercise or statins — and roughly one in five people worldwide has an elevated level.

The case for treating it was unusually strong. Multiple genetic and observational studies established Lp(a) as an independent, apparently causal risk factor for atherosclerotic cardiovascular disease. People who inherit variants giving them lifelong high Lp(a) have more heart attacks. That kind of genetic evidence is normally considered among the most reliable available, because inherited variants are not confounded by lifestyle.

And unlike LDL cholesterol, which has several effective drug classes, there was no approved therapy to lower Lp(a). A large untreated population with a genetically validated target is about as attractive as cardiovascular drug development gets.

The trial

Lp(a)HORIZON enrolled more than 8,300 patients with elevated Lp(a) who had established cardiovascular disease or a prior cardiovascular event. Pelacarsen, an antisense drug discovered by Ionis Pharmaceuticals and licensed to Novartis in February 2019, was tested against placebo.

The primary endpoint was a composite of cardiovascular death, non-fatal myocardial infarction, non-fatal stroke, and urgent coronary revascularisation requiring hospitalisation.

Importantly, patients were receiving guideline-directed therapy throughout — lipid-lowering and antihypertensive treatment, with mean LDL cholesterol around 66 mg/dL. This was a well-treated population.

The result

Pelacarsen lowered Lp(a) as intended. It produced no reduction in cardiovascular events versus placebo. Full details are expected around the American Heart Association Scientific Sessions in November.

The reaction among cardiologists was pointed. “So treating a non-causal marker of disease doesn’t treat the disease? Shocking,” one wrote — sarcasm carrying a serious argument about how much weight the field had placed on genetic causality.

Four explanations, and nobody yet knows which

The reduction was insufficient. Lifelong genetic elevation may require a far larger or longer-sustained drop than a trial can deliver in a few years.

The patients were treated too late. Lp(a) exerts its effect across a lifetime. Intervening after established disease may come after the damage is done — the same argument made about anti-amyloid drugs in Alzheimer’s.

The background therapy left no room. With mean LDL at 66 mg/dL and guideline-directed treatment throughout, residual risk attributable to Lp(a) may have been too small to detect.

Lp(a) is not directly causal. The uncomfortable possibility: that genetic association reflects something correlated with Lp(a) rather than the particle itself doing the damage.

What it means for the other programmes

Several Lp(a)-lowering drugs remain in late-stage development — among them olpasiran, lepodisiran and zerlasiran, none yet approved. Their readouts have become considerably more consequential.

The distinguishing question is depth of lowering. Some of these agents are RNA interference therapies reported to reduce Lp(a) far more profoundly than antisense approaches. If insufficient reduction explains pelacarsen’s failure, a deeper-lowering drug could still succeed. If the problem is that Lp(a) is not causal, or that intervention comes too late in established disease, they will fail the same way — and the field will have run several very expensive trials to learn it.

What to tell someone who already knows their Lp(a) is high

There is an awkward clinical consequence. European guidance has recommended that every adult have Lp(a) measured at least once in their lifetime, on the reasoning that a genetically fixed risk factor only needs checking once and identifies people who warrant more aggressive management. Testing has accordingly become more common, and a substantial number of people now know they carry an elevated level.

Until this readout, the implicit answer to “what do I do about it?” was: control everything else aggressively, and wait — a treatment is coming. The first attempt has now failed.

The practical advice does not actually change, because it was never Lp(a)-specific: lower LDL cholesterol hard, control blood pressure, do not smoke. But the expectation of a targeted therapy has moved from near-term to uncertain, and one cardiologist reacting to the result went as far as questioning whether universal Lp(a) measurement still makes sense if nothing can be done with the number. That is a minority position, and premature while other agents are still reading out — but it is now a live argument rather than an unthinkable one.

The recurring lesson

This is the second prominent cardiovascular outcomes trial of its kind to fail in the same manner: a drug engaging its target cleanly, moving its biomarker convincingly, and leaving hard clinical endpoints untouched. Ziltivekimab did it with inflammation and hsCRP; pelacarsen has now done it with Lp(a).

Both were built on strong mechanistic and epidemiological foundations, and both are reminders that a risk marker — even a genetically validated one — is not automatically a treatment target. It is exactly why large outcomes trials remain necessary, and why surrogate endpoints should be treated as hypotheses rather than conclusions. Clinical and business news, not medical or investment advice.