Novartis shed close to 4% in Zurich on Monday after pelacarsen, its lipoprotein(a) lowering drug, missed the primary endpoint of a Phase III cardiovascular outcomes study. The number that should interest investors more belongs to Amgen, which fell about 5% against Friday’s US close on a readout from a molecule it does not own.
Coverage since Friday has settled on one frame: a Novartis drug failed. That is accurate and close to useless, because it folds two very different findings into one. Pelacarsen lowered Lp(a) substantially and still did not prevent heart attacks, strokes or cardiovascular deaths. Either the target itself is wrong, which would mean the roughly 17,700 patients enrolled across Amgen’s and Eli Lilly’s ongoing outcomes trials are testing a dead hypothesis, or pelacarsen did not lower Lp(a) far enough, which would mean those trials are the ones that finally test it properly. Almost nobody covering the print has separated those two cases, and every dollar of the repricing depends on which one is true.
The 72% Problem
Pelacarsen is an antisense oligonucleotide given as a monthly injection. Across prior studies it lowered Lp(a) by an average of about 72%, and Novartis confirmed that lower Lp(a) levels were again achieved in Lp(a)HORIZON, in a population already receiving guideline-directed lipid-lowering and antihypertensive therapy. The drug did what it was designed to do to the biomarker. It just did not move the outcome.
That 72% is the whole argument. The competing molecules are not in the same range. As Fierce Biotech reported, analysts at William Blair flagged the arithmetic immediately: rival RNA interference candidates cut Lp(a) by more than 90%, so if benefit only appears past some threshold of suppression, a drug that stops at 72% produces exactly the result Novartis just reported while saying nothing about drugs that go deeper.
Citi’s Geoff Meacham was careful to leave the question open, telling clients that full data are needed to determine whether the miss reflects pelacarsen’s mechanism of action, inadequate Lp(a) reduction, trial design, or genuine doubt that lowering Lp(a) reduces cardiac risk at all.
“The first dedicated outcomes failure lowers confidence across the class and places greater pressure on later studies to demonstrate that deeper lowering of Lp(a) can produce a clinically meaningful reduction of major cardiovascular events.”
Four programs now carry that pressure, and they are not equivalent:
- Pelacarsen (Novartis and Ionis), antisense, roughly 72% reduction. Missed its outcomes endpoint.
- Olpasiran (Amgen), siRNA, above 95% at higher Phase 2 doses. OCEAN(a)-Outcomes enrolled 7,297 patients.
- Lepodisiran (Lilly), siRNA, above 90%.
- Muvalaplin (Lilly), the only oral candidate, which lowered Lp(a) by roughly 85% in its Phase 2 KRAKEN study. MOVE-Lp(a) is enrolling 10,450 patients.
Why the Market Marked Down Molecules That Have Not Reported
Amgen did not report anything on Friday. Neither did Lilly. Both moved anyway, because a first-in-class outcomes failure strips out the option value that lets investors treat an unproven mechanism as probably fine. Zurich Cantonal Bank went furthest, writing on Monday that pelacarsen simply does not work and pulling up to $1.3 billion of risk-adjusted revenue out of its Novartis model.
That is a defensible call on Novartis. It is a much harder call to defend on Amgen. The read-across trade prices a mechanism verdict the trial did not deliver, on the strength of a molecule that was, on the one variable in question, the weakest in the class. Investors have effectively concluded that if 72% did nothing, 95% will also do nothing. That is not how dose-response has behaved elsewhere in lipid biology, where the statin and PCSK9 literature points consistently the other way: more lowering, more benefit.
Here is where we land. The class sell-off is an overreaction to Novartis’s dose and a fair reaction to Novartis’s judgment. The failure is real and the company owns it, but not for the reason the tape suggests. Novartis and Ionis committed thousands of patients and years of enrollment to an outcomes trial run at a level of Lp(a) suppression they could observe was shallower than what competitors were achieving in mid-stage work. That was a capital allocation decision rather than bad luck. Novartis shareholders are paying for that decision. Amgen shareholders are paying for something they had no part in.
What Tuesday Prices In
US equity markets were shut on Monday for Labor Day, so the American leg of this repricing has not happened yet. Amgen’s 5% is an indicative move against Friday’s close, and the cash market reopens Tuesday at 9:30 a.m. Eastern with three days of accumulated read-across to clear in a single session.
The thing worth watching is not the size of Amgen’s move but whether it holds once the full Lp(a)HORIZON dataset appears at an upcoming medical congress. As BioPharma Dive noted, the two larger siRNA programs now carry most of the burden of testing the Lp(a) hypothesis. If the detailed data show benefit tracking with depth of suppression, or a signal in the subpopulation with the highest baseline Lp(a), the read-across unwinds and Amgen’s decline was a gift. If the curve is flat no matter how far Lp(a) falls, Meacham’s most pessimistic branch is live and Lilly’s 10,450-patient oral program becomes the most expensive open question in cardiovascular medicine.
Novartis still has remibrutinib and del-desiran inside a growth story investors were told is worth more than $10 billion in peak annual sales. One of the three legs just came off. The company will spend the next several months arguing that pelacarsen was a molecule problem rather than a target problem. Its competitors, who need that argument to be true, will be making it for free.