Victims of Our Own Success
A CardioAdvocate Insight — The Lp(a) Story After HORIZON
The Promise
For more than a decade, lipoprotein(a) — Lp(a) — has been one of the most intellectually compelling targets in preventive cardiology. The case for its causality was built on the strongest kind of evidence we have: human genetics.
Mendelian randomization studies — a form of natural experiment that uses genetic variants as "instruments" to test whether a biomarker truly causes disease — have consistently demonstrated that lifelong elevations in Lp(a) are causally associated with atherosclerotic cardiovascular disease and calcific aortic stenosis. The genetic architecture is unusually clean: Lp(a) levels are 75–90% heritable, determined almost entirely by the LPA gene locus. People born with genetically high Lp(a) have more heart attacks. People born with genetically low Lp(a) have fewer. This isn't confounded the way observational epidemiology can be — it's as close to a randomized experiment as nature gets.
So the reasoning was straightforward: if Lp(a) causes the disease, and we can now lower it by 80–95% with targeted therapies, then lowering it should reduce events. Multiple drug companies bet billions on this logic. The field waited for the outcomes trials.
On September 4, 2026, the first answer arrived — and it was not the one anyone hoped for.
HORIZON: What Happened
Lp(a)HORIZON — Pelacarsen Phase 3 CVOT
Drug: Pelacarsen — an antisense oligonucleotide (ASO) targeting apolipoprotein(a) mRNA production
Population: 8,323 patients with established cardiovascular disease and Lp(a) ≥70 mg/dL, all on guideline-directed medical therapy
Pharmacodynamic result: ~80% reduction in Lp(a), as expected from phase 2 data
Primary endpoint (MACE): Not met — no significant reduction in the composite of cardiovascular death, nonfatal MI, nonfatal stroke, or urgent coronary revascularization, in either the overall population or the pre-specified ≥90 mg/dL subgroup
Safety: Acceptable; no new signals
Source: Ionis/Novartis topline results announcement, Sep 4, 2026; STAT News; HCPLive
The drug worked pharmacologically. It lowered Lp(a) by approximately 80%, exactly as it did in phase 2. The biological plumbing functioned. But the clinical endpoint — fewer heart attacks, fewer strokes, fewer deaths — did not follow.
This is the first cardiovascular outcomes trial ever completed for a dedicated Lp(a)-lowering therapy. And it is unambiguously negative.
Does HORIZON Kill the Lp(a) Hypothesis?
No. But it forces the field to confront a deeper question: why did the genetics say one thing and the trial say another?
Lipidologist Pablo Corral, MD offered one of the most thoughtful post-HORIZON analyses, identifying several reasons why this single trial should not be over-extrapolated. His key points, and ours, center on a phrase that keeps coming back: victims of our own success.
1. The Background Therapy Problem
HORIZON enrolled patients who were already on intensive secondary prevention therapy. Their median LDL-C was approximately 65 mg/dL — an achievement that would have been unthinkable two decades ago. They were on statins, many on ezetimibe, some on PCSK9 inhibitors. They were on antiplatelet therapy, blood pressure medications, and comprehensive risk-factor management.
In other words, these patients were already receiving the best secondary prevention the world has ever offered. The residual risk left over — the slice of risk that Lp(a) lowering could theoretically carve into — may have been narrower than anyone anticipated when the trial was designed.
2. Lifelong Exposure vs. Late Intervention
Mendelian randomization captures the effect of a lifetime of elevated Lp(a) — from birth through decades of cumulative arterial exposure. HORIZON, by contrast, intervened in patients who already had established atherosclerotic disease — patients whose arteries had already been exposed to elevated Lp(a) for 50, 60, or 70 years. Lowering Lp(a) after the damage is done may be fundamentally different from never having high Lp(a) in the first place.
We see this pattern elsewhere in cardiovascular medicine. Cumulative LDL exposure — sometimes called "LDL-years" — is what drives atherosclerotic burden. A statin started at age 30 has profoundly different effects than one started at age 65 with three-vessel disease. The genetics tell you what happens over a lifetime. The trial tells you what happens over 3–5 years in the last chapter.
3. Lower-Than-Expected Event Rates
Because the background therapy was so good, the event rates in HORIZON were likely lower than the trial was originally powered to detect a difference against. When fewer events occur overall, it becomes statistically harder to demonstrate that the experimental drug is responsible for a meaningful portion of the reduction. This is not a flaw in the drug — it's a testament to how effective modern secondary prevention has become.
4. Lp(a) Concentration vs. Pathogenic Cargo
This is a subtler point, but potentially the most important. Lp(a) is not just a cholesterol-carrying particle. Its most atherogenic properties may come from the oxidized phospholipids (OxPL) on its surface and the prothrombotic properties of its apolipoprotein(a) tail. Two patients with identical Lp(a) levels may carry very different amounts of OxPL — and therefore very different biological risk. If the trial selected patients by concentration (Lp(a) ≥70 mg/dL) rather than by biological potency, it may have diluted the treatment effect by including patients whose Lp(a) was numerically high but biologically less dangerous.
5. The Enrollment Challenge: High, but Not the Highest
HORIZON enrolled patients with Lp(a) ≥70 mg/dL (approximately ≥175 nmol/L), with a pre-specified subgroup at ≥90 mg/dL. This captures a broad population with "elevated" Lp(a) — but it may not have concentrated on the patients at the very highest risk.
Patients with Lp(a) above 300 nmol/L (>120 mg/dL) — the top 1–2% of the population — carry the most extreme Lp(a)-attributable risk. These are the patients with strong family histories of premature heart disease, recurrent events despite optimal therapy, and the clearest unmet need. A trial that enrolls broadly across the ≥70 mg/dL population may not have been enriched enough in the highest-risk patients to detect a signal.
We await the full subgroup analyses, particularly in the highest Lp(a) strata and in analyses of total and recurrent events (rather than just time-to-first-event), which may yet reveal pockets of benefit.
The Evolution of Lipid-Lowering Therapy — And Why It Makes Trials Harder
To appreciate HORIZON's result, you need to understand how far lipid-lowering therapy has come — and how that very progress creates a paradox for clinical trials.
In the 1990s, secondary prevention trials were conducted against a backdrop of modest LDL-C lowering. The 4S trial (1994) randomized patients with LDL-C averaging 188 mg/dL to simvastatin or placebo. In that environment, statin therapy produced a dramatic 30% reduction in total mortality. There was so much residual risk to modify that the signal was unmistakable.
Fast-forward three decades. Today's secondary prevention patients are on high-intensity statins, often with ezetimibe, sometimes with PCSK9 inhibitors or inclisiran. Achieved LDL-C levels of 50–70 mg/dL are routine in well-managed patients. Antiplatelets, ACE inhibitors, beta-blockers, and aggressive blood pressure control further reduce event rates. The baseline risk against which any new therapy must prove itself has shrunk dramatically.
This is a triumph. But it creates a painful irony for drug development: the more successful we are at treating heart disease, the harder it becomes to prove that the next therapy provides incremental benefit. Residual risk reduction requires either enormous trials, very long follow-up, or very high-risk populations — or all three.
The Silver Lining: We're Doing a Bang-Up Job
Here is the part of this story that gets lost in the disappointment: if HORIZON failed partly because the patients' background therapy was so good, then we should acknowledge what that means. Contemporary secondary prevention works. Aggressive lipid-lowering therapy, antiplatelet agents, blood pressure control, and metabolic optimization are keeping people alive who, a generation ago, would have suffered recurrent events and died.
The 2026 ACC/AHA/NLA Dyslipidemia Guidelines codify this progress: tiered LDL-C targets (<55 mg/dL for very-high-risk ASCVD, <70 for standard secondary prevention), apolipoprotein B as a co-target, five new lipid-lowering therapies since 2018, and universal Lp(a) screening as a Class 1 recommendation. The toolkit is deeper than it has ever been.
If we are honest with our patients, the HORIZON result carries a bittersweet but important message: we may not yet have a drug specifically for your Lp(a), but the secondary prevention you are already on is doing more to protect you than anything available even ten years ago.
The Unmet Need: Patients We Desperately Need to Help
But the silver lining is cold comfort for a specific group of patients — and they are the ones who haunt every lipidologist's conscience.
Consider: a 42-year-old woman whose father died of a heart attack at age 48. Her LDL-C is 95 mg/dL — not terrible. She exercises, doesn't smoke, eats well. Her coronary artery calcium is zero. By every conventional risk calculator, she's "low risk." But her Lp(a) is 425 nmol/L — in the top 1% of the population. Her father's Lp(a) was never checked.
What do we tell her? That the genetics say her risk is real, but the first drug to try to address it failed its trial? That we can lower her LDL-C further with a statin — which she doesn't want to take because her LDL isn't that high? That there's nothing specific we can do about her Lp(a) right now except wait and hope?
This is the population that needs Lp(a)-lowering therapy the most: patients with very high Lp(a) (>300 nmol/L), strong family histories, and either no established disease yet (primary prevention) or early, aggressive disease that strikes despite apparently adequate risk-factor control. These patients carry an excess risk that is almost entirely genetic, almost entirely Lp(a)-driven, and almost entirely untreatable by current means.
The fact that HORIZON's failure may have been specific to its population (patients with established disease on intensive background therapy) does not reduce the urgency for this group. If anything, it redirects the question: should we be testing Lp(a)-lowering therapies in primary prevention, where the cumulative exposure argument is stronger and the background therapy is lighter?
VESALIUS-CV: The Case for Aggressive Primary Prevention
If there's a trial that illuminates why HORIZON's failure might be about timing rather than biology, it's VESALIUS-CV.
Presented at AHA 2025 and published in the New England Journal of Medicine, VESALIUS-CV randomized 12,257 high-risk patients without prior MI or stroke to evolocumab (a PCSK9 inhibitor) or placebo on top of optimized statin therapy. After a median follow-up of 4.6 years, evolocumab reduced 3-point MACE by 25%.
The median LDL-C achieved in the evolocumab arm was 45 mg/dL.
This is a landmark number. It suggests that in high-risk primary prevention — patients who have not yet had a heart attack or stroke but are at high risk for one — driving LDL-C to levels previously reserved for secondary prevention produces substantial, measurable benefit. A prespecified subanalysis of 3,655 primary prevention participants without significant atherosclerosis reinforced this: VESALIUS extends "lower is better" into territory we had not previously tested with this rigor.
What does this have to do with Lp(a)? Everything. VESALIUS demonstrates that intervening before disease is established — when the arterial wall hasn't yet accumulated decades of injury — produces a cleaner signal. If the Lp(a) hypothesis is right, it may be that the place to test it is in primary prevention, not in patients whose coronary arteries are already riddled with plaque. VESALIUS also suggests that a treatment goal of LDL-C 45 mg/dL may become the new benchmark for high-risk primary prevention, further raising the bar for what constitutes adequate background therapy — and further narrowing the window for any additional agent to prove incremental benefit.
Obicetrapib: An Unexpected Contender
In a field reeling from HORIZON's disappointment, one of the more intriguing developments comes from an unexpected direction: obicetrapib, a next-generation CETP inhibitor developed by NewAmsterdam Pharma and its Chief Scientific Officer, Dr. John Kastelein.
CETP inhibitors have a troubled history. Torcetrapib raised blood pressure and increased mortality. Dalcetrapib had no efficacy. Evacetrapib was neutral. Anacetrapib showed a modest benefit but was abandoned by Merck for commercial reasons. Many assumed the CETP class was dead.
Obicetrapib is different — by design. It achieves near-complete CETP blockade without the off-target effects that plagued torcetrapib (no RAAS activation, no blood pressure increase). And its lipid profile is striking:
- LDL-C reduction: 33–41% on top of maximally tolerated statin therapy, in the Phase 3 BROADWAY and BROOKLYN trials
- Lp(a) reduction: ~45% — modest compared to dedicated Lp(a)-lowering agents, but meaningful as an add-on effect
- ApoB reduction: Significant, reflecting fewer atherogenic particles overall
- Small, dense LDL particles: Reductions of up to 54–102%, addressing the most atherogenic subfraction of LDL
But what has Dr. Kastelein and others truly excited goes beyond lipids.
Pleiotropic Effects: Beyond the Lipid Panel
Diabetes risk: CETP inhibitors as a class have been associated with reduced risk of new-onset diabetes and improved insulin sensitivity — the opposite of what statins do. In an era of cardiometabolic medicine where diabetes and cardiovascular disease are increasingly recognized as one syndrome, a lipid-lowering drug that improves glycemic parameters is noteworthy.
Alzheimer's disease biomarkers: In a remarkable finding, analysis of 1,727 patients from the obicetrapib Phase 3 program showed that obicetrapib significantly attenuated the increase in plasma p-tau217 and the p-tau217/Aβ42:Aβ40 ratio — key biomarkers of Alzheimer's pathology. Mendelian randomization studies have separately suggested that lower CETP activity is associated with reduced dementia risk. The mechanism likely involves CETP's role in brain lipid metabolism — the brain is a cholesterol-dependent organ, and the lipoproteins that shuttle cholesterol across the blood-brain barrier may be modulated by CETP activity.
Brain lipids and small particles: CETP regulates the exchange of cholesterol esters and triglycerides between HDL and LDL particles. By blocking this exchange, obicetrapib reduces the number of small, dense LDL particles — the most atherogenic subfraction — while increasing the cholesterol content of larger, less harmful particles. This particle-size shift may have implications beyond the arterial wall, including in the central nervous system where lipid transport is critical for neuronal health.
What Comes Next for Lp(a)?
HORIZON is a setback, not a verdict. Three critical data points remain:
| Trial | Drug (Class) | Lp(a) Reduction | Expected Results |
|---|---|---|---|
| OCEAN(a)-Outcomes | Olpasiran (siRNA, Amgen) | ~94–95% | Late 2026 |
| ACCLAIM-Lp(a) | Lepodisiran (siRNA, Eli Lilly) | ~94% | 2027–2028 |
| PREVAIL | Obicetrapib (CETP-i, NewAmsterdam) | ~45% (+ LDL-C ↓40%) | Late 2026 |
The siRNA trials (olpasiran and lepodisiran) are mechanistically distinct from pelacarsen: they achieve even larger Lp(a) reductions (>90% vs. ~80%) through RNA interference rather than antisense oligonucleotide-mediated mRNA degradation. If HORIZON failed because 80% Lp(a) lowering wasn't enough, siRNAs may clear that bar. If it failed because of the population, the background therapy, or the timing — then even 95% reduction may not be enough in the same setting.
ACCLAIM-Lp(a), notably, enrolls both secondary prevention and high-risk primary prevention patients (16,700 total) — a design that may answer the question HORIZON could not: does Lp(a) lowering work better when you intervene earlier?
What HORIZON Does Not Change
It is critical to be precise about what HORIZON does and does not tell us:
- Lp(a) is still causal. The Mendelian randomization evidence — from large GWAS studies, from the AHA Scientific Statement, from the EAS Consensus Statement — has not been invalidated. One pharmacological trial does not override the genetics.
- Screening still matters. The 2026 ACC/AHA/NLA Dyslipidemia Guidelines recommend measuring Lp(a) at least once in every adult (Class 1, LOE B-NR). This stands. Knowing your Lp(a) sharpens your risk assessment, informs cascade screening of family members, and drives more aggressive management of modifiable risk factors.
- PCSK9 inhibitors remain Class 1 for patients with ASCVD and elevated Lp(a) not at LDL-C goal. The FOURIER and ODYSSEY OUTCOMES data showing greater benefit in patients with higher Lp(a) are independent of HORIZON.
- The aortic stenosis question is separate. Lp(a)'s causal role in calcific aortic valve disease is established through distinct Mendelian randomization analyses. HORIZON tested only ASCVD endpoints; whether Lp(a) lowering prevents aortic stenosis remains an untested and important hypothesis.
An Honest Conversation
If you have elevated Lp(a), here is what we can now say with intellectual honesty:
Your Lp(a) is a real risk factor. It's genetic. It's causal. Knowing your number is important because it changes how aggressively we manage everything else — your LDL-C target, your blood pressure, your metabolic health, your lifestyle. That has not changed.
What has changed is the timeline for a dedicated Lp(a)-lowering drug. A year ago, many of us expected pelacarsen to be the first FDA-approved Lp(a) therapy, potentially available within 1–2 years. That is very unlikely now. The field's hopes rest on olpasiran (late 2026) and lepodisiran (2027–2028), but the bar of evidence these drugs must clear has been raised considerably by HORIZON's failure.
In the meantime, the best thing you can do is what the 2026 guidelines say: treat everything you can treat. Get your LDL-C to goal. Optimize blood pressure. Address metabolic health. Exercise. Don't smoke. If you have ASCVD and elevated Lp(a) and your LDL-C isn't at target, a PCSK9 inhibitor is indicated — and it will lower your Lp(a) modestly (~25–30%) as a secondary benefit.
For patients with very high Lp(a) (>300 nmol/L) and progressive disease despite optimal therapy, Lp(a) apheresis remains FDA-approved and available. For everyone else: cascade screening your family. Make sure your siblings, parents, and children know their Lp(a). The conversation doesn't end because one drug trial was negative.
The Bottom Line
We are victims of our own success. The revolution in secondary prevention — high-intensity statins, combination lipid-lowering therapy, PCSK9 inhibitors, and aggressive risk-factor management — has compressed residual cardiovascular risk to the point where proving that any new agent provides incremental benefit has become extraordinarily difficult.
The Lp(a)HORIZON trial failed not because the hypothesis is wrong, but potentially because we tested it in a setting where success was hardest to demonstrate: in patients whose other risks were already well-controlled, whose arteries had already sustained decades of damage, and whose event rates were lower than anticipated because modern therapy works.
Lp(a) is still causal. Screening is still essential. Aggressive management of modifiable risk factors in the presence of elevated Lp(a) is still guideline-directed. The genetics haven't changed.
What has changed is our humility. We now know that lowering a causal biomarker doesn't guarantee clinical benefit in every context. We may need to intervene earlier, in different populations, with different trial designs. VESALIUS-CV shows us that primary prevention with aggressive lipid lowering (LDL-C to 45 mg/dL) works. Obicetrapib reminds us that the future of lipid-lowering may be broader than we imagined — a single oral pill that touches LDL-C, Lp(a), small particles, diabetes risk, and possibly even Alzheimer's pathology.
The story of Lp(a) is not over. But its next chapter will be written by different drugs, in different patients, with different expectations. And the silver lining, if we're honest enough to see it, is that the reason this chapter is so hard is that we're already doing so well.
Don't let HORIZON's result talk you out of knowing your number. Know it. Share it with your family. And fight like hell on every risk factor you can fight.
CardioAdvocate Rounds
The Weekly Dispatch for Heart-Smart Patients
Every Friday, we distill the week's most important cardiometabolic developments into a focused newsletter — so you stay informed without the noise.
Subscribe to RoundsFree • Weekly • No spam
