For years, the cleanest untested idea in cardiology was that lipoprotein(a), an inherited risk factor carried by roughly one in five people, could be treated like cholesterol: lower it, and cardiovascular events fall. On September 4, Novartis reported that the first large trial to test that idea in outcomes failed. Pelacarsen, an antisense drug that reduces Lp(a) at its source, did not meet its primary endpoint in the Lp(a)HORIZON study of 8,323 patients with elevated Lp(a) and established cardiovascular disease.
The failure does not settle the hypothesis. It reshapes it. The company has disclosed no hazard ratio, no effect size, and no subgroup numbers, saying full data will come at a medical congress. What the market heard in Zurich on Monday was the fact of the miss, and Novartis shares fell more than 3 percent, erasing roughly 4.9 billion Swiss francs of value.
The trial that was supposed to settle it
Lp(a)HORIZON was the field's flagship. It randomized patients with Lp(a) levels of 70 milligrams per deciliter or higher, a threshold that marks the inherited condition, who already had cardiovascular disease and were on standard care. The primary endpoint was the four-part composite the field uses: cardiovascular death, nonfatal heart attack, nonfatal stroke, and urgent coronary revascularization requiring hospitalization. The trial assessed the overall population and a prespecified subgroup with Lp(a) of 90 or higher.
Pelacarsen is a monthly injectable antisense oligonucleotide that Novartis licensed from Ionis in 2019, and its earlier data were exactly what a hypothesis needs: phase 2 studies showed Lp(a) reductions of roughly 72 to 80 percent. The missing link was always outcomes. Lowering a risk marker does not by itself prevent events, a lesson the field learned from HDL, and Lp(a)HORIZON was the experiment designed to prove the connection. It did not.
Novartis chief medical officer Shreeram Aradhye called the results "not the results we hoped for, but they provide important evidence that advances scientific understanding," and the company's statement leaves the program's future open, with language about possible future submission rather than a discontinuation decision. For the roughly 20 percent of people who carry elevated Lp(a) with no approved targeted therapy, the practical outcome is unchanged: standard cardiovascular risk management remains the only option, and the targeted drugs remain experimental.
The read-across lands on the rivals
The failure immediately re-priced the rest of the class. Amgen's olpasiran, in its own outcomes trial with 7,297 patients, is the closest competitor, and it suppresses Lp(a) more deeply than pelacarsen did, above 95 percent. Eli Lilly has lepodisiran and an oral candidate, muvalaplin, in a 10,450-patient trial. Silence Therapeutics has earlier-stage programs. The analyst argument that kept those programs valuable was that deeper suppression might matter where 72 to 80 percent did not, and Citi's Geoff Meacham summarized the damage in one line: "The first dedicated outcomes failure lowers confidence across the class." Amgen and Lilly shares moved down in the same session, though U.S. markets were closed for the Labor Day holiday and the repricing will complete over the following sessions.
The failure lands on Novartis at an awkward moment. Days earlier, the company suspended a cell therapy trial after three patient deaths, and analysts now point to its remibrutinib program in multiple sclerosis as the near-term catalyst, with Jefferies suggesting that win "should make the conclusion of the pelacarsen study more palatable." Peak-sales forecasts for pelacarsen had ranged from $1.5 billion at UBS to $5.4 billion at Jefferies, and at least one Swiss bank has already stripped its estimate from the model entirely. Ionis, which was entitled to royalties in the mid-teens to low-twenties percent, saw its shares fall harder than Novartis's.
What survives the failure
The scientific questions the trial leaves open are specific. The event rate in the placebo arm, the actual Lp(a) reduction achieved in a real-world population, and the 90-plus subgroup results are all unknown until full disclosure. If the subgroup showed a signal, the theory narrows but survives. If the reduction was smaller than the phase 2 data promised, the drug failed for a dosing reason rather than a mechanism reason. Either way, the class now needs a different kind of proof than the one it just lost.
There is also a harder possibility that the trial forces the field to confront. Lp(a) may be a marker of risk rather than a driver of it, the way decades of failed HDL-raising drugs eventually forced cholesterol research to distinguish the two. The genetics that identified Lp(a) as causal are strong, and Mendelian evidence is why the drugs were built. But genetics measure lifelong exposure from birth. A trial of a few years in patients who already have disease tests whether reversing the marker late helps, which is a different question. The honest state of the field is that the first answer came back negative, and the second question is now the one that matters.
For patients, nothing changes yet
For people with elevated Lp(a), the failure changes nothing about their care today, and that is worth saying plainly. No targeted drug was approved before the trial, and none is approved now. The standard recommendations, controlling LDL cholesterol, blood pressure and other modifiable risks, remain the evidence-backed path, and the trial's failure underlines rather than weakens that guidance. People enrolled in ongoing trials should continue them and discuss any concerns with their care team.
The Lp(a) hypothesis did not die in Zurich this week. It lost its first big test, which is what first tests are for, and the next two years will tell whether the class finds its proof in deeper suppression, in a narrower population, or in a different endpoint. The one outcome that is no longer available is the easy one.
Primary sources
- Novartis's Lp(a)HORIZON topline announcement for the trial design, endpoint, and company statement.
- Reuters reporting, syndicated by MarketScreener, for the market reaction and rival read-across.
- dpa-AFX reporting, syndicated by ARIVA, for the Swiss banking commentary.
- Clinical Trial Vanguard for the drug's mechanism and prior Lp(a) reduction data.
- BusinessTech for the analyst discussion of dose and class implications.
- Il Sole 24 Ore for the cell therapy trial suspension context.