Why Lowering Lp(a) Did Not Cut Events in 8,323 Patients
Novartis said pelacarsen lowered Lp(a) in patients already on guideline-directed care, but Lp(a)HORIZON missed its cardiovascular composite versus placebo. The first large outcomes test of the inherited risk factor now hangs over later trials from Amgen and Lilly.
On September 4, 2026, Novartis announced that pelacarsen missed the primary endpoint of Lp(a)HORIZON, the first large cardiovascular outcomes trial of a drug built to lower lipoprotein(a). The Phase 3 study randomized 8,323 patients with established cardiovascular disease and screening Lp(a) of at least 70 mg/dL to 80 mg of the antisense oligonucleotide injected monthly or matching placebo. The composite was cardiovascular death, non-fatal myocardial infarction, non-fatal stroke, and urgent coronary revascularization requiring hospitalization. Lower Lp(a) levels were achieved on top of guideline-directed lipid lowering and antihypertensive therapy. That reduction did not produce fewer events in the overall study population.
Shreeram Aradhye, Novartis's president of development and chief medical officer, said the findings "did not demonstrate that this translated into reduced cardiovascular risk in the overall study population." He called HORIZON "a pioneering cardiovascular outcomes trial designed to answer one of the most important unanswered questions in cardiovascular medicine." Ionis Pharmaceuticals, reported the same miss, adding that pelacarsen produced substantially lower Lp(a) levels consistent with previous studies and an acceptable safety profile. Full data will be presented at an upcoming medical congress. Neither company disclosed a hazard ratio, p-value, event counts, or the Phase 3 percent reduction in Lp(a).
Residual risk, already squeezed
Lp(a)HORIZON was designed as a randomized, placebo-controlled, double-blind, multicenter trial in adults 18 to 80 with prior myocardial infarction, ischemic stroke, or symptomatic peripheral artery disease. A co-primary analysis in patients with Lp(a) of at least 90 mg/dL was built into the protocol. Novartis has not separately disclosed that stratum.
The American Heart Journal design paper describes 8,323 patients randomized at 797 sites in 42 countries. Mean age was 59.7 years, and 27.0% were women. Qualifying events were myocardial infarction in 81.5%, stroke in 9.8%, and peripheral artery disease in 14.2%. High-intensity statin use was 77.5%, ezetimibe or bile acid binding resins 56.6%, and a PCSK9 inhibitor 10.8%. Median baseline LDL-C was 64.6 mg/dL (interquartile range 52.0 to 80.8). Median baseline Lp(a) was 108.2 mg/dL, or 235.7 nmol/L. In all, 78.6% (6,540 patients) sat at or above 90 mg/dL. The trial was event-driven until 993 confirmed primary endpoint events, with more than 90% power to detect a hazard ratio of about 0.80 in the full population or 0.75 in the 90 mg/dL subpopulation, assuming a 4.6% annual placebo event rate.
Mendelian randomization cited in that paper estimated that an absolute Lp(a) reduction of 70 to 90 mg/dL might be needed for roughly 15 to 20% risk reduction, similar to a 1 mmol/L drop in LDL-C, which is why the enrollment floors were set at 70 and 90 mg/dL. Pharmacodynamic modeling indicated 80 mg monthly should produce about an 80% time-averaged Lp(a) reduction, similar to 20 mg weekly in Phase 2b. That 80% figure is not a reported Phase 3 result. In the 2020 New England Journal of Medicine Phase 2 study of 286 patients, mean percent Lp(a) reductions from baseline at six months ranged from 35% at 20 mg every four weeks to 80% at 20 mg weekly, compared with a 6% reduction on pooled placebo.
Novartis exercised its option in February 2019, taking worldwide development from Akcea, an Ionis affiliate, and paid $150 million, split evenly between Ionis and Akcea, while assuming Phase 3 costs. Brett P. Monia, Ionis's chief executive, said the company was disappointed that substantial Lp(a) lowering "did not translate to cardiovascular risk reduction." Novartis has not said it is discontinuing pelacarsen. No approval filing has been made.
What the miss does not settle
Genetic and epidemiologic evidence had treated Lp(a) as causal, about 90% inherited and largely untouched by diet or lifestyle. HORIZON is the first large pharmacologic test of that claim, and it missed in the overall population even though on-treatment Lp(a) fell and background LDL-C was already low. Open questions include whether residual risk after optimized LDL-C was too small for this sample, whether the 70 mg/dL floor sat below the Mendelian-randomization burden, whether antisense lowering differs from later small interfering RNA programs, and what the undisclosed 90 mg/dL stratum will show. Guidelines still recommend once-in-a-lifetime Lp(a) testing. Elevated Lp(a) still affects about one in five people worldwide, with no approved targeted treatment. The miss undercuts using Lp(a) lowering as a surrogate for approval in secondary prevention until full data, and later trials, report.
Two outcomes programs have not read out. Amgen's olpasiran trial, OCEAN(a) Outcomes, has enrolled 7,297 patients with Lp(a) of at least 200 nmol/L and atherosclerotic disease, dosed every 12 weeks, with estimated primary completion of March 31, 2028. HORIZON's median baseline was 235.7 nmol/L, so OCEAN(a)'s 200 nmol/L floor sits nearer HORIZON's middle than its 70 mg/dL entry bar. Lilly's lepodisiran trial, ACCLAIM-Lp(a), is estimated at 17,300 patients with Lp(a) of at least 175 nmol/L and either established disease or age of at least 55 at risk for a first event, with primary completion estimated in March 2029. Those design differences are the live tension, not a forecast that either trial will succeed. Until they report, Lp(a) remains a risk factor clinicians can measure and a target no drug has yet been shown to treat.