Same Mechanism, Different Trial: How the FDA Rejected Patisiran but Approved Vutrisiran

First, the disease. Transthyretin (TTR) is a protein the liver makes. In ATTR amyloidosis it misfolds and deposits as amyloid fibrils in tissue; when those deposits build up in the heart muscle, they stiffen its walls and drive a progressive, ultimately fatal heart failure — cardiac ATTR amyloidosis, or ATTR-CM. The same protein can instead settle in peripheral nerves, the hereditary polyneuropathy form (hATTR-PN).

Patisiran and vutrisiran attack that disease the same way — RNA interference, targeted to liver cells, shutting down TTR production at the source. Patisiran is an IV infusion every three weeks in a lipid nanoparticle; vutrisiran is a subcutaneous shot every three months. Mechanistically they are nearly interchangeable — which is what makes the pair a clean experiment: hold the molecule constant, and the only thing that changed was the trial.

Why the FDA rejected patisiran — APOLLO-B and the CRL

Patisiran's cardiac trial, APOLLO-B, was built short and simple: 360 patients, 12 months, a single primary endpoint — the six-minute walk distance, a measure of how far a patient can walk on flat ground. Patisiran hit it: a 14.7-meter gain in favor of the drug (p=0.0162). By its own design's standards, the trial succeeded.

What it could not show in twelve months was a hard cardiovascular outcome. Its secondary composite of death and cardiovascular events — the kind of endpoint that actually matters in heart failure — came back at a win ratio of 1.27 (p=0.0574), favoring patisiran but short of statistical significance: too few events, in too short a trial, to land it. So when the FDA's advisory committee voted 9-3 that the benefit supported approval, the agency overrode the panel and issued a Complete Response Letter — no safety or manufacturing problem, simply that the effect was small and its clinical meaningfulness unestablished. A walk-distance gain at twelve months was not, in the cardiology reviewers' judgment, proof on a disease that kills through heart failure.

The bar APOLLO-B fell short of was not written for amyloidosis. It was set by a decade of large cardiovascular trials built on death and hospitalization measured over years, not a walk test over months. Even within ATTR cardiomyopathy the precedent was explicit: tafamidis, the one prior approval, had won on a hierarchical mortality-and-hospitalization composite, with the six-minute walk only a secondary measure. (The walk test is not weak in the abstract — it cleared a generation of pulmonary-hypertension drugs — but ATTR cardiology had already set a higher bar.) Patisiran's trial inverted that ranking.

How vutrisiran won FDA approval — HELIOS-B's hard-outcome trial

Vutrisiran's trial, HELIOS-B, was the opposite shape. It enrolled 655 patients and followed them up to 36 months against a hard composite of all-cause mortality and recurrent cardiovascular events — scored by a win ratio, which ranks death above hospitalization and lets a patient's repeat events count instead of stopping at the first. Vutrisiran cut that composite by 28%, with a 36% reduction in all-cause mortality. In March 2025 the FDA approved it for cardiomyopathy — and, on the same label, for the polyneuropathy form: the first drug to carry both.

One caveat sits under that headline. About 40% of HELIOS-B's patients were already on tafamidis, which APOLLO-B's were not; the cleanest signal came from the tafamidis-naïve group, a pre-specified co-primary population (hazard ratio 0.67). The standard of care was not identical between the two trials. But the axes this story turns on — duration, endpoint, statistical power — all moved the same way.

Alnylam has never said it redesigned HELIOS-B because APOLLO-B failed, but the trial was plainly built to meet the bar APOLLO-B missed. And the obvious objection — that vutrisiran simply won because it is the better molecule — does not survive contact with the chemistry. The two differ only in delivery, a quarterly injection versus a three-weekly infusion, not in what they do to the heart; both knock down transthyretin by more than 80%. On identical biology, vutrisiran's win suggests patisiran could have cleared a hard-outcome trial too — but it was never given one.

It was the trial, not the molecule

That this was never really about the molecule is clearest from a third drug. In the same window, BridgeBio's acoramidis — a small-molecule stabilizer, an entirely different mechanism — won ATTR cardiomyopathy on the same kind of long, hard-outcome, win-ratio trial. Alnylam invented the RNAi modality and reached ATTR first; on the cardiac indication — the one that mattered commercially — it finished third, behind tafamidis and acoramidis, with its own successor drug.

And the bar is not moving down. The next ATTR-CM trials — eplontersen's CARDIO-TTRansform and Intellia and Regeneron's in-vivo CRISPR therapy NTLA-2001 — are being built long and hard from the first protocol. For a one-time, irreversible gene edit, that matters most of all: what will stand between it and approval is the same thing that decided patisiran's fate — a trial built to clear the bar.

Step back, and the lesson isn't really about amyloidosis. A drug's fate at the FDA can be sealed long before the first patient is dosed — in how long the trial runs, how many it enrolls, who those patients are, and what it sets out to measure. Patisiran got the molecule right and the trial wrong, and the molecule lost.

This story turned on one piece of trial design: what you measure — the endpoint. There are probably others where the design mattered even more. One I keep coming back to turns on a different lever: who you enroll. AstraZeneca's gefitinib failed an all-comers lung-cancer trial (ISEL, 2005) and was all but pulled from the U.S. market, then was reinstated a decade later, paired with a companion diagnostic, once a new trial enrolled only the EGFR-mutation patients in whom it worked. The molecule never changed; the patients did. It's the same lesson as patisiran's, reached through a different part of the trial — and there are others.