Novo Nordisk Halts Ziltivekimab Cardiovascular Programme After Futility Review

Novo Nordisk has stopped the two remaining heart-failure outcomes trials of its ziltivekimab cardiovascular programme after a data-monitoring committee judged them unlikely to differ from the null ZEUS result. The decision leaves one post-infarction study standing and tightens the evidentiary bar for residual-inflammation claims in market-access dossiers.

Investigators learned on 4 September 2026 that HERMES and ATHENA would close early. The company confirmed the move on 7 September after an independent data-monitoring committee reviewed the totality of the evidence and concluded there was a low likelihood that either heart-failure study would produce a different result from ZEUS, the Phase 3 outcomes trial that missed its primary endpoint in July.

The molecule did what it was designed to do on the bench and in the blood. It did not do what payers, HTA bodies, and cardiology guidelines require in the clinic. That gap, target engagement without event reduction, now has to be written into any residual-inflammation value narrative.

Ad

Ziltivekimab cardiovascular programme: what stopped and what remains

Ziltivekimab is a fully human monoclonal antibody directed at the interleukin-6 ligand. Novo Nordisk acquired it with Corvidia Therapeutics in 2020 for 725 million US dollars upfront and up to 2.1 billion dollars in milestones. Consensus peak sales before ZEUS sat near 3 billion dollars. The asset was the company’s most visible attempt to build a cardiovascular franchise that did not depend on GLP-1 agonists.

The late-stage programme had four named pillars:

Patients in the stopped trials will be followed to their scheduled three-month visits. ARTEMIS remains on the original plan. That study tests a different inflammatory setting, the weeks after infarction, rather than chronic residual risk in stable atherosclerotic and kidney disease.

ZEUS: biology moved, events did not

ZEUS was a multinational, double-blind, placebo-controlled, event-driven trial. Participants had established atherosclerotic cardiovascular disease, chronic kidney disease, and high-sensitivity C-reactive protein of at least 2 mg/L. They were randomised to ziltivekimab 15 mg subcutaneously once monthly or matching placebo, on top of standard of care.

Headline results, released 31 July 2026:

• Primary 3-point MACE (cardiovascular death, non-fatal myocardial infarction, or non-fatal stroke): hazard ratio 0.99, with a 95% confidence interval of 0.88 to 1.11.
• Free interleukin-6 and hsCRP fell as expected. The antibody engaged its target.
• All-cause mortality did not differ between arms.
• Overall adverse-event rates were similar. Serious infections were more common with ziltivekimab, consistent with other interleukin-6 pathway drugs.

Martin Holst Lange, Novo Nordisk’s chief scientific officer, said the trial “provides important scientific evidence” but that ziltivekimab “did not achieve the MACE benefit we had hoped for.” Shares fell in the high single digits on the day. The company flagged a non-cash impairment in the third quarter and left its 2026 adjusted operating-profit outlook unchanged.

The CKD inclusion criterion was not incidental. Low-dose colchicine is the only anti-inflammatory with a US cardiovascular indication in chronic stable atherosclerosis, and it is contraindicated in clinically important kidney disease. ZEUS was built to occupy that gap. The null result removes the cleanest regulatory and access rationale the programme had.

The hypothesis, not a single molecule, is under pressure

The inflammatory hypothesis of atherothrombosis has never rested on one antibody. It rests on a chain: residual hsCRP after lipid control predicts events; innate-immunity pathways drive plaque; blocking those pathways should cut MACE even when LDL is already low. CANTOS (canakinumab, interleukin-1β) supplied the first human outcomes signal in 2017, with a 15 percent relative reduction at the 150 mg dose and no lipid effect. Canakinumab was not taken forward for a cardiovascular label, largely because of cost and an excess of fatal infections.

Subsequent tests have been uneven.

Lowering an inflammatory biomarker is not the same as preventing an event. ZEUS and CLEAR SYNERGY both moved CRP and both missed their clinical composites. Setting matters, too. A signal in stable coronary disease or in a narrowly selected post-infarction window has not automatically transferred to kidney-disease cohorts, heart-failure cohorts, or unselected post-MI populations on contemporary background therapy.

That last point is why ARTEMIS is still running. Acute infarction has a different cytokine profile from the chronic, statin-treated, high-CRP phenotype enrolled in ZEUS. Whether that distinction is biologically real or a rationale for keeping one option open will be known in the first half of 2027.

This week’s two cardiovascular misses are different

The same news cycle carried Novartis and Ionis’s announcement that pelacarsen missed the primary four-point MACE composite in Lp(a)HORIZON (8,323 patients with elevated lipoprotein(a) and established cardiovascular disease). Pelacarsen lowered Lp(a) and did not reduce events. Some coverage folded both failures into a single “inflammatory-disease hypothesis” story. They are not the same hypothesis.

Pelacarsen tests whether genetically anchored residual lipid risk, measured as Lp(a), is a treatable driver of events once LDL is controlled. Ziltivekimab tests whether residual innate-immunity risk, measured as hsCRP and interleukin-6, is a treatable driver of events once lipids and other standards of care are in place. Both programmes show the same pattern, the biomarker moved and the composite did not, but the access and modelling implications sit in different dossiers. Amgen and Eli Lilly still have competing Lp(a)-lowering assets. Those programmes should be scored on Lp(a) reduction, trial design, and background therapy, not on ZEUS.

What this means for value, HTA, and market access

For an access leader the operational question is not whether inflammation “exists” in atherosclerosis. Genetic, pathologic, and epidemiologic evidence for that is not erased by one antibody. The question is whether a residual-inflammation claim can still carry incremental price, a restricted pathway, or a place in a cost-effectiveness model.

• Surrogate-to-outcome chains are now weaker. Dossiers that treat a fall in hsCRP or interleukin-6 as a reliable proxy for MACE reduction will face harder HTA questioning. ZEUS is a clean counter-example: expected biomarker movement, hazard ratio of 1.
• The CKD gap is unfilled. Colchicine’s contraindication in important kidney disease was the practical opening for an interleukin-6 antibody. That opening has not been converted into an outcomes benefit. Payers should not build an IL-6 CKD rider into cardiovascular pathway models on current evidence.
• Heart-failure inflammation claims are not ready for submission planning. HERMES and ATHENA were the registration-intent tests of that idea. A futility stop after ZEUS is not a peer-reviewed heart-failure result, but it is enough to freeze budget-impact work that assumed a 2027-2028 HFpEF launch.
• ARTEMIS is a different decision problem. If it is positive, the label conversation will be acute-care and secondary-prevention, not chronic residual risk in CKD. Pricing, site of care, and the comparator set would all change. If it is null, the late-stage IL-6 cardiovascular file is effectively closed.
• Portfolio strategy at Novo Nordisk tightens. The halt further concentrates growth on the obesity and diabetes franchise at a moment when that franchise already faces Eli Lilly. External access teams should expect more business-development activity in cardiovascular and rare disease, not a near-term IL-6 launch sequence.

HTA agencies that have been asked to consider inflammatory biomarkers as effect modifiers in cardiovascular models now have a large, contemporary outcomes trial that failed to convert those markers into events. That does not prohibit subgroup analyses. It does mean any such analysis needs a pre-specified clinical composite, not a CRP story.