gift link: https://www.nytimes.com/2026/09/04/science/heart-drug-fails-...

Damn, biology is complex.

But I'm glad our system at least has enough integrity to admit failures of this magnitude.

>The Novartis study lasted seven years and involved 8,323 people who had high Lp(a) levels and pre-existing heart disease.

>The C.E.O. of Pfizer, Dr. Krumholz said, “thought the whole future of his company would be built on this drug.”

This result should be kept in mind by all those techbros saying they're going to fix the drug discovery business, make it efficient and get things cranking out in 18 months. Sometimes, the dog dies. Sometimes, the model isn't quite close enough to the problem. Sometimes, they find there's a second mechanism that reveals itself at the wrong time.

Goddamn, this is a disappointment. Everyone thought (and probably still thinks) Lp(a) was/is causal. There is a very "neat" story about how the shorter LPA genes (with fewer Kringle repeats) leaves the Lp(a) cholesterol particle's surface exposed, and the shorter genes over-expressed their products. Although I don't think anyone demonstrated the specific damage mechanism, it sure seemed like that could be a smoking gun.

That said, I look forward to seeing the actual data. Since they enrolled patients with previous coronary events, the damage may have already been done, and lowering Lp(a) levels might not do much in that case. Still, it was hoped that the damage would stop accumulating, leading to noticeably fewer events.

Beautiful theories once again murdered by ugly facts.