AHA 2026 — Conference Coverage
When the Biomarker Lies
The Billion-Dollar Lesson from AHA 2026
This article is for educational purposes only and does not constitute medical advice. Clinical trial interpretation is nuanced — the conclusions drawn here reflect the published data as of November 2026. Always consult your healthcare provider regarding medication decisions and treatment plans.
Image: AI-generated for CardioAdvocate
The Case
It's November 2026. You're watching the American Heart Association Scientific Sessions. Two trials — HORIZON and ZEUS — publish simultaneously in the New England Journal of Medicine. Both are large, well-conducted, Phase 3 randomized controlled trials. Both hit their biomarker targets with precision. And both failed to move the needle on the only thing that matters: heart attacks, strokes, and death.
Sound familiar? It should. Cardiology has been here before.
The Pattern
In 1989, the Cardiac Arrhythmia Suppression Trial (CAST) shook the foundations of evidence-based medicine. Encainide and flecainide were brilliant at suppressing premature ventricular complexes — the abnormal heart beats that everyone "knew" predicted sudden cardiac death. The drugs worked perfectly on the EKG monitor. And they killed patients. Mortality went up, not down.
The lesson was supposed to be permanent: improving a surrogate marker is not the same as improving patient outcomes.
Thirty-seven years later, we learned it again — multiple times, at the same conference, in the same week.
HORIZON: The Lp(a) Chapter
Lipoprotein(a) — Lp(a) — has been the great hope of preventive cardiology. Decades of Mendelian randomization data showed that people born with genetically high Lp(a) have higher rates of cardiovascular disease. The genetics were as clean as it gets in epidemiology: no confounders, no reverse causation, just a straightforward signal that high Lp(a) causes atherosclerosis.
Pelacarsen, an antisense oligonucleotide from Novartis and Ionis, lowered Lp(a) by approximately 78% in 8,323 patients with established cardiovascular disease. The primary MACE endpoint was 12.3% in the treatment group versus 12.1% with placebo. Hazard ratio: 1.02 (95% CI 0.90–1.15, P = 0.76). Median follow-up: 4.7 years.
The biomarker was crushed. The patients weren't helped.
ZEUS: The Inflammation Chapter
Inflammation has been the other great hope. After CANTOS showed that blocking IL-1β with canakinumab modestly reduced cardiovascular events — and colchicine showed promise in COLCOT and LoDoCo2 — the next logical step was upstream: block IL-6 itself, the master cytokine that drives CRP and the entire acute-phase response.
Ziltivekimab, Novo Nordisk's anti-IL-6 antibody, was tested in 6,376 patients with atherosclerotic cardiovascular disease, chronic kidney disease, and elevated hsCRP (≥2 mg/L) in the ZEUS trial. The drug lowered IL-6 by 76% and CRP by 64%. The MACE hazard ratio was 0.99 (95% CI 0.88–1.11, P = 0.84). Not a trend. Not a hint. Nothing.
Worse, ziltivekimab carried real safety costs: higher rates of serious infections (20.4% vs 17.9%), venous thromboembolism (1.6% vs 0.6%), and thrombocytopenia (7.4% vs 3.2%). All risk, no reward.
Novo Nordisk subsequently halted the HERMES (heart failure) and ATHENA trials of ziltivekimab. The IL-6 program is over.
The CAST Lesson, Updated
| Year | Trial | Biomarker Target | Biomarker Result | Patient Outcome | Off-Target Effects? |
|---|---|---|---|---|---|
| 1989 | CAST | PVCs (arrhythmia suppression) | Suppressed | Mortality ↑ (harm) | Yes — pro-arrhythmic |
| 2006 | ILLUMINATE | HDL (raised with torcetrapib) | Raised 72% | Death ↑ (harm) | Yes — raised blood pressure, aldosterone |
| 2014 | AIM-HIGH | HDL/Triglycerides (niacin) | Improved | No benefit | Flushing, glucose ↑ |
| 2026 | HORIZON | Lp(a) (lowered with pelacarsen) | ↓78% | No benefit | None observed |
| 2026 | ZEUS | IL-6/CRP (lowered with ziltivekimab) | IL-6 ↓76%, CRP ↓64% | No benefit | Yes — infections, VTE, thrombocytopenia |
The pattern is unmistakable: modulating a biomarker — even one with strong biological rationale and clean genetic epidemiology — does not guarantee clinical benefit.
An Important Distinction
The CAST analogy is instructive — but context matters. The CAST drugs were pro-arrhythmic: they had dangerous off-target effects that directly killed patients, independent of whether they suppressed PVCs. The ILLUMINATE torcetrapib story was similar — the drug raised blood pressure and aldosterone through an off-target mechanism, causing harm beyond what raising HDL could offset. In those cases, the drugs did not just fail to help; they actively harmed through pharmacological effects unrelated to the biomarker.
The HORIZON result is fundamentally different. Pelacarsen did exactly what it was designed to do — lower Lp(a) dramatically — and no serious off-target effects emerged. The drug appears safe. It simply didn't produce the cardiovascular benefit we expected from lowering the biomarker. That is a failure of the biomarker hypothesis, not a failure of drug safety.
ZEUS sits somewhere in between: ziltivekimab's safety signals (infections, thromboembolism) represent real off-target harm, more reminiscent of CAST and ILLUMINATE. So the comparison to CAST is appropriate — to a degree. These are all cautionary tales about chasing biomarkers. But the why each trial failed differs, and that matters for what comes next.
The Oldest Biomarker Tells the Same Story
Lp(a) and IL-6 may be the biomarkers of the moment, but the oldest version of this lesson involves the most familiar number in medicine: blood pressure.
As we wrote in our Under Pressure deep dive: "Blood pressure is a biomarker. Hypertension is a disease." That distinction sounds academic — until you look at what happens when you lower the number without asking why the number is high and who you're treating.
In 2010, the ACCORD-BP trial randomized 4,733 patients with type 2 diabetes to intensive blood pressure control (systolic <120 mmHg) versus standard control (<140 mmHg). The number dropped. The composite of heart attack, stroke, and cardiovascular death did not — HR 0.88 (95% CI 0.73–1.06, P = 0.20). Intensive therapy did reduce stroke rates, but at the cost of more adverse events, more medications, and no overall cardiovascular benefit.
Five years later, the SPRINT trial tested the same strategy — same targets, same biomarker — in 9,361 high-risk patients without diabetes. This time it worked: HR 0.75 (95% CI 0.64–0.89) for the primary composite, plus a significant reduction in death. The trial was stopped early for benefit.
Same biomarker. Same target. Opposite results. The difference was the disease, not the number.
Now, at AHA 2026 — the very same conference that brought us HORIZON and ZEUS — this story repeated itself with remarkable precision:
| Trial (Year) | Population | BP Target | CV Outcome |
|---|---|---|---|
| ACCORD-BP (2010) | Type 2 diabetes, high CV risk | <120 vs <140 | No benefit (HR 0.88, P = 0.20) |
| SPRINT (2015) | High CV risk, no diabetes | <120 vs <140 | Benefit (HR 0.75, stopped early) |
| OPTIMAL-DIABETES (2026) | Type 2 diabetes, high CV risk | <120 vs <140 | No benefit |
| OPTIMAL-STROKE (2026) | Prior ischemic stroke/TIA | <120 vs <140 | Benefit (10.9% vs 13.5%) |
OPTIMAL-STROKE enrolled 4,368 patients with prior ischemic stroke or TIA at 30 sites in Brazil. Over 3.9 years, targeting systolic <120 mmHg reduced the composite of cardiovascular death, MI, stroke, and hospitalization for unstable angina or heart failure compared to <140 mmHg. OPTIMAL-DIABETES enrolled 9,476 patients with type 2 diabetes — a larger trial — and found no significant benefit with the same targets.
Sixteen years apart, the pattern held: aggressive blood pressure lowering works in some high-risk populations and fails in others. Diabetes, with its complex web of metabolic, microvascular, and inflammatory pathology, resists the simple logic of "lower the number, fix the patient." The disease defines whether the biomarker intervention translates to outcomes — not the biomarker itself.
As we noted in Under Pressure: "It's one thing to link a biomarker to adverse health outcomes, but it's much harder to develop interventions that, when reducing that biomarker, lead to a lowering of those adverse events." HORIZON, ZEUS, ACCORD-BP, and now OPTIMAL-DIABETES all learned this the hard way.
Why This Matters Beyond Cardiology
This lesson extends far beyond the clinic. Consider what millions of Americans do every day:
- Take nattokinase because it shows "fibrinolytic activity" in a test tube — a biomarker — while replacing their prescribed anticoagulant
- Take berberine because small studies show modest glucose reductions — a biomarker — instead of proven diabetes medications
- Inject BPC-157 — a "street peptide" sold online with a "not for human consumption" label — because rodent studies show tissue repair markers
- Take turmeric/curcumin because it "reduces inflammation" in short-term studies — a biomarker — as a substitute for evidence-based anti-inflammatory therapies
If pelacarsen, a $1 billion drug developed over a decade and tested in 8,000+ patients, can crush its biomarker target and fail to help a single patient, what should we conclude about supplements and "street peptides" whose evidence amounts to test-tube studies and rat experiments?
The answer is uncomfortable: the biomarker story is not the patient story. Moving a number on a lab report — Lp(a), CRP, IL-6, fasting glucose, fibrinolytic markers — is the beginning of the evidence journey, not the end.
Related Reading
The Wild, Wild West: Supplements & Heart Health — The unregulated world of supplements, separating evidence from marketing in the $60 billion industry
Peptides: Wild West or New Frontier? — Street peptides, compounding pharmacies, and the underground market built on biomarker promises
Victims of Our Own Success — Why modern medicine's greatest triumphs make the next breakthrough harder to prove
Under Pressure: What You Need to Know About Hypertension — Why blood pressure is a biomarker, hypertension is a disease, and the distinction matters
What AHA 2026 Does NOT Change
Before the nihilists declare victory, let's be clear about what still stands:
1. Lp(a) is still a real risk factor. HORIZON's failure does not mean Lp(a) doesn't cause disease. It means pelacarsen, given to patients with established CVD on aggressive background therapy, didn't reduce it enough — or in the right patients — or at the right time. Three more Lp(a) outcome trials (olpasiran, lepodisiran, muvalaplin) are underway.
2. Inflammation still matters in atherosclerosis. CANTOS showed a signal. Colchicine shows consistent benefit across two positive and one neutral trial. ZEUS tells us that systemic IL-6 blockade is the wrong approach, not that inflammation is irrelevant.
3. Statins still work. LDL lowering with statins, ezetimibe, PCSK9 inhibitors, and bempedoic acid remains the cornerstone of cardiovascular prevention. These drugs passed the outcomes test — the test that HORIZON and ZEUS failed.
4. Lp(a) screening is still recommended. The 2026 ACC/AHA/NLA guidelines give Lp(a) testing a Class 1 recommendation. Knowing your Lp(a) helps refine risk — even if we can't lower it with a drug yet.
THE BOTTOM LINE
AHA 2026 will be remembered as the conference that taught cardiology — again — to respect the gap between biomarkers and outcomes. CAST taught us in 1989. ILLUMINATE reinforced it in 2006. ACCORD-BP showed us in 2010 that even the most familiar biomarker can lie in the wrong population. HORIZON, ZEUS, and OPTIMAL-DIABETES delivered the lesson again in 2026 — while SPRINT and OPTIMAL-STROKE reminded us that the same biomarker can tell the truth when the biology is right.
For clinicians: keep using proven therapies. Don't chase biomarkers that haven't been validated with outcomes data.
For patients: when someone — a supplement company, a gym trainer, a wellness influencer, or a gray-market peptide seller — tells you their product "improves your markers," ask the follow-up question: "Has it been proven to reduce heart attacks, strokes, or death?" If the answer is no, you're betting on a biomarker. And November 2026 just showed us how that bet can end.
CardioAdvocate is not anti-innovation. We are pro-evidence. The pipeline continues. The science is not done. But the standard of proof remains: show us the outcomes.
— CardioAdvocate
CardioAdvocate provides medical education content for informational purposes only. This article does not constitute medical advice. Clinical trial data and guideline recommendations are subject to revision as new evidence emerges. Always consult your healthcare provider regarding medication decisions. CardioAdvocate has no financial relationship with Novartis, Ionis Pharmaceuticals, Novo Nordisk, or any pharmaceutical company mentioned in this article. © 2026 CardioAdvocate. All rights reserved.
