A single infusion cut LDL cholesterol 62% by rewriting one letter of DNA in the liver

Gene editing has so far reached patients mostly by leaving the body first. Cells are removed, edited in a facility, and returned. The alternative — infusing the editing machinery and letting it find its target inside a living organ — has been the field's stated destination for a decade.

A phase 1 trial published on 25 May in the New England Journal of Medicine reports results from one such attempt. VERVE-102 is designed to permanently inactivate PCSK9 in liver cells after a single intravenous infusion [s1].

The rationale, which is unusually clean

PCSK9 is a well-mapped target. People who carry naturally occurring loss-of-function variants of the gene have lower LDL cholesterol and fewer atherosclerotic cardiovascular events than people without them [s1]. The genetic experiment has, in effect, already been run in the human population.

Two classes of drug already exploit this — monoclonal antibodies and RNA interference — but both require repeat dosing indefinitely. A base edit does not. It changes the gene once.

What was given

The study was a phase 1, open-label, single-ascending-dose trial in adults with heterozygous familial hypercholesterolemia or premature coronary artery disease [s1]. Participants received one intravenous infusion at one of six dose levels, ranging from 0.3 to 1.0 mg of total RNA per kilogram of body weight [s1].

The therapy itself is two RNAs in a lipid nanoparticle: a messenger RNA encoding an adenine base-editor protein, and a guide RNA targeting PCSK9 [s1]. The nanoparticle incorporates N-acetylgalactosamine, the sugar tag that liver cells take up preferentially [s1].

A total of 35 participants across the six cohorts received VERVE-102 and had at least 28 days of follow-up [s1].

What happened

The pharmacological effect was substantial and dose-dependent. Mean reductions in blood PCSK9 protein ranged from 51% at the 0.3 mg/kg dose to 88% at 1.0 mg/kg [s1].

Corresponding LDL cholesterol reductions ranged from 9% at the lowest dose to 62% at the highest, an absolute drop of 78 mg per decilitre at 1.0 mg/kg [s1]. The reductions appeared durable across the available follow-up, which was at least one year in 15 of the 35 participants [s1].

On safety, no dose-limiting toxic effects occurred [s1]. Mild-to-moderate infusion-related reactions and transient elevations in alanine aminotransferase — a liver enzyme — were observed [s1]. Aspiration pneumonitis occurred in one participant who had gastroesophageal reflux disease [s1].

What the trial does not establish

Four things, and each matters more than usual because the intervention is irreversible.

It measured cholesterol, not outcomes. The stated objectives were safety and changes in blood PCSK9 protein and LDL cholesterol levels [s1]. LDL reduction is a well-validated surrogate for cardiovascular risk, and reduction of this magnitude by other mechanisms has translated into fewer events. But no participant in this trial was followed to a heart attack endpoint, and a phase 1 single-ascending-dose study is not built to do that.

There was no control group. It was open-label and single-arm [s1]. For a biochemical endpoint that moves as far as 88% below baseline this is a smaller concern than usual, but adverse events in an uncontrolled trial have nothing to be compared against.

The dose groups are very small. Thirty-five participants spread across six dose levels leaves roughly half a dozen people per cohort [s1]. The reported reductions are cohort means from handfuls of individuals, and safety signals occurring in fewer than one in thirty-five people would not be expected to appear at all.

Follow-up is short relative to permanence. Most participants had at least 28 days; 15 had at least a year [s1]. The edit is intended to last a lifetime. Durability at one year is genuine evidence that the modification persists, and it is not evidence about years five, ten or thirty — neither for continued efficacy nor for late consequences.

The reversibility asymmetry

With an injectable antibody or an siRNA, a problem discovered in year three can be addressed by stopping the drug. With a base edit, there is nothing to stop.

That is the trade the field is proposing, and it is a reasonable one to consider for people whose inherited cholesterol levels cause coronary disease decades early. It also means the evidentiary bar for long-term safety is different in kind from that for a chronic medication, and the current dataset — 35 people, one year at most — sits well below it.

What to watch

The relevant next steps are the ones this trial was designed to enable: larger cohorts at the doses that worked, and eventually a randomised outcomes trial that measures cardiovascular events rather than lipid panels. Long-term registries will matter more here than in most drug development, because the exposure never ends.

The trial was funded by Verve Therapeutics and is registered as NCT06164730 [s1].