Amgen Slid 10% on a Rival's Failed Trial. The Real Question Is Whether That Was a Failed Drug or a Dead Idea

A 10% haircut landed on Amgen on Tuesday, but the trial behind it wasn't Amgen's. On September 4, Novartis announced that pelacarsen, its drug for the cholesterol particle called Lp(a), had failed a cardiovascular outcomes study. The drug lowered Lp(a) levels in more than 8,000 patients, yet did not reduce heart attacks, strokes, or cardiovascular deaths. Wall Street took that miss as a verdict on the idea itself: that driving down Lp(a) keeps people alive. That is the same idea underneath Amgen's own Lp(a) drug, olpasiran.

By midday, Amgen had shed about 10% to roughly $394 a share — about $44, or more than $20 billion of market value, from a pharmaceutical company that makes no money on olpasiran at all. The slide carried the Dow with it, and Amgen's decline ran far deeper than Eli Lilly's, which fell only about 2% on the same news.

The failed step is the whole point

Pelacarsen worked exactly as designed and still failed. In its 8,323-patient Phase 3 trial, the drug cut Lp(a) by roughly two-thirds to three-quarters in prior studies — before, in the outcomes trial, its lower levels simply did not translate into fewer cardiovascular events on top of what patients were already getting. Patients in these studies are not untreated; they arrive on optimized statins and other guideline-directed therapy. The trial asked whether knocking out Lp(a), on top of all that care, removes the residual risk the particle carries. The answer Novartis got was no.

That failure lands hardest on the layer of the chain that never gets a headline: the biological premise itself. Lp(a) is a fat-and-protein particle, roughly 90% genetically determined, that silently raises risk of plaques and rupture in perhaps a fifth of the population. It has long been treated as a mandated patient — risk that must be lowered. Pelacarsen was meant to prove the mandatory step: cut the particle, cut the events. When the largest test of that step comes back empty, every drug waiting on the same premise is collateral.

Amgen's olpasiran is the largest of those stragglers.

Two ways to read one failure

Whether Amgen's slide is an overreaction or a fair correction turns on a single distinction: was Novartis's miss a failed drug, or a failed idea?

The drug-versus-idea gap matters, because olpasiran is not pelacarsen. Mechanics differ. Pelacarsen is an antisense oligonucleotide; olpasiran is a small interfering RNA, a different way of switching off the liver's production of the protein. The dose effect differs more sharply: olpasiran has reduced Lp(a) by more than 95% at week 36 at effective doses, where pelacarsen managed roughly two-thirds to 80%. If the hypothesis is that near-elimination of the particle is required for benefit, olpasiran is a genuinely different bet and the Novartis result is the wrong yardstick.

But there is a darker reading, and it is the one analysts at Citi flagged. If the failure is not about how much Lp(a) a drug removes, but about whether removing Lp(a) on top of modern care changes outcomes at all, then a deeper knockdown does not rescue Amgen's molecule — it just makes a more dramatic version of the same futility. In that case Novartis's trial, whatever its design flaws or dose gaps, has weakened the hypothesis that every Lp(a) program was priced on. You cannot know which reading is true from the topline; that distinction sits in sub-analyses, in Eli Lilly's differently-engineered Lp(a) program, and eventually in olpasiran's own readout.

What the market just priced out

The most useful number in Tuesday's move is not the 10%. It is what $44 a share represents: a discount on a drug that has no revenue, no filing, and no answer coming soon.

Olpasiran's Phase 3 outcomes trial, OCEAN(a)-Outcomes, has enrolled roughly 7,300 patients with established cardiovascular disease and elevated Lp(a). It began in December 2022 and is not expected to complete until around March 2028. The primary endpoint is the strip of events that matters — coronary heart disease death, heart attack, and urgent coronary revascularization. In other words, the market is not pricing earnings; it is pricing a two-years-off bet on whether a hypothesis holds. A $20 billion-plus swing on an option that far away tells you how much of Amgen's value was already standing on olpasiran's peak case.

Here is the purity test that usually separates a tollbooth from a theme stock, turned on its head. Amgen is not a pure play on anything. It runs roughly a $200 billion-market-cap portfolio whose near-term earnings ride on Repatha, the LDL-lowering injectable that grew 34% year over year to $876 million in the first quarter, on oncology drugs, and on its still-unproven obesity franchise. Olpasiran is optionality bolted to the side of that machine — a high-purity bet on one hypothesis, but zero current revenue. A shareholder who bought Amgen for Repatha and the pipeline, treating olpasiran as free upside, was not materially wronged by Tuesday. A shareholder who bought it for olpasiran just watched the odds on that one bet collapse.

The market's own behavior draws the line. The same day it sold Amgen on Lp(a), Amgen posted a positive Phase 3 result — better overall survival in small cell lung cancer for its drug IMDELLTRA — and the stock still fell. That is an unusual split: the drop was asset-specific, aimed squarely at the Lp(a) option, not at the health of the company. The market was able to separate the two; so should the reader.

Know which reason you owned it

The practical consequence of Tuesday is not a new valuation. It is a forced clarity about why you were holding or watching Amgen at all. If the answer was the near-term business — lipid markets, oncology, obesity — then Novartis's failure changed the evidence you were trading on only at the margin, and a double-digit drop is opportunity noise. If the answer was olpasiran's blockbuster potential, then the read-through is legitimate: the largest test of your hypothesis came back negative, and no amount of clever mechanism hand-waving overrides that until olpasiran's own data arrive.

The clock runs until roughly 2028. The confirmation metric for the Lp(a) bull case is any follow-up evidence — cardiology analyses of pelacarsen's failure, or Lilly's differently-built program — showing a subgroup or mechanism that still converts Lp(a) reduction into few events. The signal that olpasiran becomes ordinary, rather than a recovered hidden winner, is the opposite: evidence that the failure was premise-level, which would leave a deeper Lp(a) knockdown as a firmer version of a dead idea. Tuesday's 10% was a market deciding, in one day, which camp it stood in. Amgen's shareholders now have to decide the same thing, only with two years of open questions ahead of them instead of one day.