One infusion of a CRISPR therapy cut ANGPTL3 levels by about 80% in 15 people

A single intravenous infusion of a CRISPR-Cas9 therapy aimed at the liver reduced blood levels of the protein ANGPTL3 by around 80% at the two highest doses tested, according to a phase 1 trial published in the New England Journal of Medicine on 8 November [s1]. Fifteen people received the drug. That is the whole trial, and the size of it is the first thing to hold on to.

Why ANGPTL3

ANGPTL3 — angiopoietin-like protein 3 — inhibits lipoprotein and endothelial lipases, the enzymes that clear fats from the bloodstream [s1]. People who happen to carry loss-of-function variants in the ANGPTL3 gene have lower low-density lipoprotein cholesterol and lower triglycerides, and a lower lifetime risk of atherosclerotic cardiovascular disease [s1]. That is an unusually clean natural experiment: a group of people whose bodies already run the intervention, apparently without penalty.

Drug developers have been trying to copy it. Evinacumab, a monoclonal antibody against ANGPTL3, was tested in a phase 2 trial of 272 patients with refractory hypercholesterolaemia — people whose LDL cholesterol stayed high despite lipid-lowering therapy at maximum tolerated doses [s2]. At week 16, subcutaneous evinacumab reduced LDL cholesterol relative to placebo by 56.0, 52.9 and 38.5 percentage points at the three subcutaneous dose regimens tested, and by 50.5 percentage points at the higher of two intravenous doses [s2]. Serious adverse events during the treatment period ranged from 3% to 16% across trial groups [s2].

The antibody approach works, in other words, but it has to be given again and again. The premise of CTX310 is different: edit the gene once and the liver stops making the protein.

What CTX310 is, and what the trial did

CTX310 is a lipid nanoparticle carrying CRISPR-Cas9 messenger RNA and a guide RNA directed at hepatic ANGPTL3, intended to induce a loss-of-function mutation — deliberately breaking the gene in liver cells [s1].

The trial was an ascending-dose phase 1 study in adults with uncontrolled hypercholesterolaemia, hypertriglyceridaemia, or mixed dyslipidaemia who were already receiving maximally tolerated lipid-lowering therapy [s1]. Each participant received a single intravenous dose at one of five levels: 0.1, 0.3, 0.6, 0.7 or 0.8 mg per kilogram of body weight [s1].

The primary endpoint was adverse events, including dose-limiting toxic effects [s1]. This is worth stating plainly, because it governs what the trial can and cannot tell anyone. A phase 1 dose-escalation study asks whether a dose can be given at all. It is not designed to show that a treatment prevents heart attacks, and this one does not attempt to.

The results

Fifteen participants received CTX310 and had at least 60 days of follow-up [s1].

No dose-limiting toxic effects related to CTX310 occurred [s1]. Serious adverse events occurred in two participants, 13% of the group: one had a spinal disk herniation, and one died suddenly 179 days after treatment at the lowest dose, 0.1 mg per kilogram [s1]. Infusion-related reactions were reported in three participants (20%) [s1]. One participant (7%), who had elevated aminotransferase levels at baseline, had a transient rise in those liver enzymes to between three and five times baseline, peaking on day 4 and returning to baseline by day 14 [s1].

The effect on ANGPTL3 tracked dose, and did so steeply. The mean percent change in ANGPTL3 level was 9.6% at 0.1 mg per kilogram (range −21.8 to 71.2), 9.4% at 0.3 mg per kilogram (range −25.0 to 63.9), −32.7% at 0.6 mg per kilogram (range −51.4 to −19.4), −79.7% at 0.7 mg per kilogram (range −86.8 to −72.5) and −73.2% at 0.8 mg per kilogram (range −89.0 to −66.9) [s1].

Two features of those numbers matter. The first is that the two lowest doses did nothing on average and the individual ranges span large increases as well as decreases — consistent with no real effect and ordinary biological variation. The second is that at 0.7 and 0.8 mg per kilogram, the ranges tighten considerably and sit entirely below zero. The threshold between "nothing happened" and "the gene is largely switched off" appears to fall in a narrow band of dose.

What is not here

The published abstract reports the ANGPTL3 protein levels and the safety findings. It does not report that this translated into any clinical outcome, and no such claim should be read into it [s1]. The authors' own conclusion is narrow: editing of ANGPTL3 was associated with few adverse events and resulted in reductions from baseline in ANGPTL3 levels [s1].

The sudden death of a participant 179 days after the lowest dose is reported as a serious adverse event [s1]. The trial reports no dose-limiting toxic effects related to the drug [s1]. In a 15-person study with no control group, a single death cannot be attributed to or cleared of a treatment relationship by the trial itself; it is a data point that larger and longer studies exist to interpret.

Sixty days of follow-up is the reporting floor here [s1]. Gene editing is, by design, permanent. The central open question for this class of medicine is what a deliberately broken gene does over decades, and no phase 1 trial can answer it.

The trial was funded by CRISPR Therapeutics, the drug's developer, and registered with the Australia New Zealand Clinical Trials Registry [s1].

What to watch

The useful comparison is not between CTX310 and nothing, but between one-time editing and repeated dosing of an antibody like evinacumab, which is already known to lower LDL cholesterol substantially in a hard-to-treat population [s2]. A durable single treatment would change the arithmetic of lifelong adherence — if the durability holds and the safety record survives contact with larger numbers of people. Neither is established by these results.

Sources

• [s1] Phase 1 Trial of CRISPR-Cas9 Gene Editing Targeting ANGPTL3, The New England Journal of Medicine, 8 November 2025. https://doi.org/10.1056/NEJMoa2511778
• [s2] Evinacumab in Patients with Refractory Hypercholesterolemia, The New England Journal of Medicine, 15 November 2020. https://doi.org/10.1056/NEJMoa2031049