Pelacarsen and Ziltivekimab Fail Late-Stage Trials, Ending the "Genetics Is Undefeated" Era in Cardiovascular Drug Development
核心洞察
Novartis's pelacarsen and Novo Nordisk's ziltivekimab both failed to separate from placebo in late-stage cardiovascular outcome trials despite strong genetic support for their targets.
Cardiologist Ethan Weiss said the field can no longer claim that genetics is undefeated in predicting which drugs will work in outcome studies.
Novartis shares fell 14% in a single session and Amgen dropped 9.8% as investors repriced roughly 35,000 patients still enrolled in Lp(a) outcome trials.
Two of the most genetically supported cardiovascular programs in development failed to separate from placebo in late-stage outcome trials within a single week in early September 2026, undermining the field's central de-risking assumption that human genetics reliably predicts which drugs will prevent heart attacks and strokes.
Novartis's pelacarsen, an antisense oligonucleotide that lowers lipoprotein(a), and Novo Nordisk's ziltivekimab, which targets the inflammatory protein IL-6, both missed their primary endpoints. The results prompted UCSF cardiologist Ethan Weiss, co-founder and chief scientific officer of Marea Therapeutics (搜索), to declare the end of a long-held conviction in cardiovascular drug development.
"We would say genetics is undefeated," Weiss said on The Readout Loud podcast. "That it always predicted which drugs were going to work in outcome studies. We cannot say that anymore."
Pelacarsen Hits Its Target, Patients Still Have Events
The Lp(a) case was the more painful of the two because the underlying logic appeared airtight. Roughly one in five people worldwide are estimated to carry elevated lipoprotein(a), a hereditary cholesterol-like particle linked to higher risk of heart attack and stroke. The measure is likely under-assessed. Statins do not affect it, and PCSK9 inhibitors move it only modestly.
Pelacarsen reduced Lp(a) by approximately 80% in phase two studies — the bare minimum, Weiss observed, that the field believed necessary to predict an event reduction. The phase three Lp(a)HORIZON trial enrolled 8,323 patients with established cardiovascular disease (搜索), all on guideline-directed therapy. The drug did exactly what it was designed to do: it crushed Lp(a) levels. Patients still had heart attacks, strokes, and died at the same rate as those on placebo.
The genetics had pointed to a 30% risk reduction in an outcome study. That did not materialize.
Weiss laid out three possible explanations: the drug is simply bad, the experiment was designed wrong, or the hypothesis itself was wrong. His lean is toward the last, and he identifies background therapy as the smoking gun. Going back 35 years, in patients well treated with statins and other existing tools, the risk signal from elevated Lp(a) "seems to go away or at least get significantly diminished."
"The genetics can tell you what happens with lifelong loss of function in a whole population of people," he explained. "It doesn't tell you what happens in a short-term trial in people who are on a lot of other really powerful medicines."
Weiss is not prepared to dismiss the target entirely but is downgrading its importance. Only about 20% of people have elevated Lp(a), and it is really only the extreme tail with very high elevations that carries meaningful risk. He continues to advocate testing Lp(a) as a risk marker, recounting educated patients arriving frightened by sky-high values, with physicians responding by maximizing statins and possibly adding a PCSK9 inhibitor — "because that's all we had." His guess now is that this may have been the right answer all along.
Ziltivekimab and the Inflammation Hypothesis
The Novo Nordisk failure arrived from a different angle but reached the same destination. Ziltivekimab targets IL-6, a protein involved in systemic inflammation long suspected of driving cardiovascular risk independent of cholesterol. The ZEUS study failed to hit its primary endpoint in July, and on September 4 the company scrapped two more late-stage trials, HERMES and ATHENA, after a data monitoring committee saw no reason to continue.
Weiss admitted the inflammation hypothesis had convinced him. "I actually expected a benefit," he said. In retrospect, he noted, the genetic signal for inflammation was never as robust as the one for Lp(a) — a warning sign the field collectively downplayed. His current view is that anyone who thought IL-6 would be the next cardiovascular blockbuster is "probably wrong," though he still assigns a future positive trial "a decent chance."
Market Repricing Across the Lp(a) Class
The financial reaction was immediate and broad. Novartis suffered its steepest single-day decline since March 2020, with European shares tumbling more than 9% after the del-desiran failure landed on top of the pelacarsen miss. The American depositary receipts fell 14% in one session — the worst on record. Ionis Pharmaceuticals, which discovered pelacarsen and licensed it to Novartis in 2019, slid about 10%. Amgen took a 9.8% hit in its worst session in 26 years as investors repriced olpasiran, its own Lp(a)-targeting siRNA drug now in phase three with 7,297 patients enrolled. The iShares Biotechnology ETF fell 1.8% on broad sector contagion.
The read-through is not automatic — different molecules, different mechanisms, possibly different patient populations — but the market no longer rewards biomarker reduction for its own sake. Roughly 35,000 patients remain enrolled in Lp(a) outcome trials built on the premise pelacarsen just undermined. Eli Lilly is running two: lepodisiran with 17,300 patients and muvalaplin with 10,450, though the latter blocks Lp(a) particle assembly through a different mechanism.
Weiss connected the failures directly to financing behavior. His prediction is that large companies will not be enthusiastic about running large cardiovascular outcome trials for a while, and for small-company investors it will be "almost impossible," especially because capital has many other destinations with less risk and less investment.
"This is going to pop a hole in the balloon, and it's going to be a long time before somebody wants to do something like this again," he said. "I think investors are going to run away from this."
Marea itself had been developing its lead program toward a cardiovascular endpoint and planning a large cardiovascular outcome study, then changed strategy about a year ago — a decision Weiss said he is now happy about, calling such a trial "really, really hard today."
A Broader Reckoning for Validation Shortcuts
The cardiovascular failures arrived in a season of wider turbulence for the industry's confidence mechanisms. In August and early September 2026, the FDA approved eight novel drugs, two cell and gene therapies, and one vaccine — an unusually productive streak for an agency operating without a permanent commissioner since Marty Makary's departure. Heidi Overton, nominated to take the role, has not yet had a Senate confirmation hearing scheduled.
Yet the approval productivity masks a transparency gap. Novartis paused its autoimmune CAR-T therapy trials on August 24 but disclosed the pause on September 1 citing "serious immune responses" — without mentioning three patient deaths. The deaths only surfaced after a reporter inquired, and critical details remain unknown, including when the deaths occurred and what autoimmune conditions the patients had. Patient deaths in investigator-initiated gene and cell therapy trials in China similarly emerged only after press questioning; those trials allow Chinese doctors and companies to test new medicines without sign-off from central regulators, and while the data cannot support a U.S. approval, they generate early readouts that investors use to validate investment theses.
Vykat XR (搜索), a drug for Prader-Willi syndrome (搜索) developed by Soleno Therapeutics and later acquired by Neurocrine Biosciences, has shown post-market side effects, toxicity, and deaths that were invisible in small trial populations — a structural vulnerability in rare-disease development where small sample sizes and favorable-looking benefit-risk profiles obscure signal until market scale. Prader-Willi is an inherited disease causing an insatiable drive to eat, with the brain signaling constant hunger.
The Merck (搜索) and Moderna personalized mRNA cancer vaccine provides the inverse example: a stock-moving late-stage win in melanoma (搜索), announced by press release with zero data disclosed, that slowed the return of melanoma and its spread to other parts of the body. Moderna's stock rose sharply on the news, renewing optimism around an mRNA modality that had faced backlash since COVID.
The common thread is that validation shortcuts — genetic, biomarker-based, early-signal — are failing at the precise moment they are asked to support the most expensive proof the industry runs.
Financial and Clinical Fallout for Novartis
Novartis faces the largest patent cliff in its history in 2026, with Entresto already losing U.S. exclusivity and Cosentyx exposure looming around 2029. The pelacarsen and del-desiran failures remove roughly $5 billion in modeled peak sales. Jefferies has said the company's 5-6% growth target through 2030 "is likely to be seen as unachievable" without further acquisitions, and Barclays expects the valuation premium to compress.
On the clinical side, Weiss argued both things are true: the field is clearly not treating everybody optimally and should use existing tools more aggressively, but trial evidence shows that even optimal treatment leaves patients with recurrent significant heart attacks. He wants to keep hunting novel pathways — "I'd like to be one of them" — while acknowledging the bar for convincing investors has risen sharply.
"I think there's going to be some dark days ahead in the space," he said. "I mean, that's just my honest assessment."
What to Watch
Full pelacarsen data at a future medical congress will reveal whether the miss was drug, design, or hypothesis — and whether any patient subgroup showed benefit worth salvaging. Amgen's olpasiran trial reads out in March 2028, and Eli Lilly's lepodisiran in March 2029; whether those trials proceed with investor confidence intact is now an open question. Novo Nordisk's ARTEMIS trial of ziltivekimab in acute heart attack continues, with data expected in the first half of 2027.
Also unresolved: whether the other Lp(a) programs following Novartis proceed, whether Heidi Overton's confirmation hearing is scheduled and what it signals for FDA direction, whether investigator-initiated trial frameworks in China face new scrutiny from investors and regulators, and whether the Prader-Willi post-market safety record changes how rare-disease trials are sized or monitored.