🧬 Aging Wiki

In Vivo Base Editing of PCSK9 with VERVE-102 for Hypercholesterolemia

7 min read

In Vivo Base Editing of PCSK9 with VERVE-102 for Hypercholesterolemia

TL;DR: The first peer-reviewed, full Phase 1 report of an in-vivo base-editing medicine. A single intravenous infusion of VERVE-102 — an adenine base editor (ABE) mRNA plus a PCSK9-targeting guide RNA, packaged in a GalNAc-lipid nanoparticle — produced dose-dependent reductions in blood PCSK9 (up to −88%) and LDL cholesterol (up to −62%, an absolute −78 mg/dL) at the 1.0-mg/kg dose, with effects that appeared durable through at least 12 months from one dose. No dose-limiting toxic effects, deaths, or withdrawals occurred across 35 participants.

Critical context: This is the publication the wiki’s crispr-base-editing-pcsk9 page had been explicitly flagging as the missing “full Phase 1 paper.” Two things matter equally here: (1) the result — somatic editing of one liver gene can reproduce, from a single dose, the LDL-lowering achieved by chronic PCSK9 monoclonal antibodies; and (2) the method — this is early human safety/PD evidence that an LNP-delivered, mRNA-encoded base editor is a tractable therapeutic platform, with implications well beyond PCSK9. The methodology is treated as a first-class entity at in-vivo-base-editing.

Design

FieldValue
TrialHeart-2 (NCT06164730)
Phase / designPhase 1, open-label, single-ascending-dose, non-randomized, single-arm (no formal statistical testing; this is a non-prespecified interim analysis)
SponsorVerve Therapeutics (wholly owned subsidiary of Eli Lilly)
SitesAustralia, Canada, New Zealand, United Kingdom
PopulationAdults 18–70 yr with heterozygous familial hypercholesterolemia (HeFH) and/or premature coronary artery disease (CAD; onset ≤55 yr men, ≤65 yr women); fasting LDL-C ≥70 mg/dL on maximum tolerated lipid-lowering therapy
Key exclusionsUncontrolled hypertension, inadequately controlled type 2 diabetes, ongoing PCSK9-inhibitor use
n (this interim)35 (data-cutoff 2026-02-27; planned enrollment up to 85)
InterventionSingle IV infusion (~4 h) of VERVE-102: mRNA encoding adenine base editor v8.8 + PCSK9 guide RNA, encapsulated in a GalNAc-lipid nanoparticle
Dose levels (total RNA)0.3 (n=4), 0.45 (n=6), 0.6 (n=4), 0.7 (n=8), 0.8 (n=6), 1.0 mg/kg (n=7)
PremedicationDexamethasone (day before + day of infusion) + histamine H₁- and H₂-receptor antagonists
ObservationInpatient ≥2 days; weekly through day 28; then days 60, 90, 180, 270, 365; up to 18 months at cutoff
Primary objectiveSafety (CTCAE v5.0)
Secondary objectivesPharmacokinetics; pharmacodynamics — % and absolute change from baseline in plasma PCSK9 and fasting LDL-C (time-averaged from day 28 by trapezoidal AUC)

Mechanism of the edit. The guide RNA directs the editor to a splice site at the 5′ end of intron 1 of PCSK9. An A·T → G·C conversion alters the splice site, enabling read-through to a stop codon that prevents translation and expression of PCSK9 protein in the liver. The ABE combines an engineered adenosine deaminase with a catalytically-impaired Cas9 nickase (nicks one strand rather than making a double-strand break). The GalNAc moiety adds asialoglycoprotein-receptor (ASGPR)-mediated hepatocyte uptake on top of the ApoE/LDL-receptor uptake of the lipid nanoparticle. See in-vivo-base-editing for the full platform mechanics and pcsk9 for target biology.

Baseline characteristics (N=35)

CharacteristicValue
Mean age (range)52 yr (27–66)
Male sex24 (69%)
Race — White / Asian / Black30 (86%) / 6 (17%) / 1 (3%)
Mean LDL cholesterol129 ± 42 mg/dL
Mean PCSK9455 ± 98 µg/L
Clinical status — HeFH only / HeFH + premature CAD / premature CAD only20 (57%) / 9 (26%) / 6 (17%)
Concomitant statin (high-intensity / moderate-low / none)25 (71%) / 7 (20%) / 3 (9%)
Concomitant ezetimibe15 (43%)

Results

Pharmacodynamics — dose-dependent, substantial

Time-averaged change from baseline (day 28 onward), by dose cohort:

Dose (mg/kg)nPCSK9 % change (range)LDL-C % change (range)
0.34−51 (−73 to −30)−9 (−26 to 26)
0.456−59 (−92 to −34)−44 (−57 to −30)
0.64−61 (−84 to −38)−45 (−54 to −34)
0.78−64 (−93 to −34)−33 (−70 to −14)
0.86−77 (−87 to −66)−51 (−86 to −16)
1.07−88 (−94 to −78)−62 (−79 to −45)

At 1.0 mg/kg the mean LDL cholesterol fell from 128 mg/dL (3.3 mmol/L) at baseline to a time-averaged 51 mg/dL (1.3 mmol/L) — an absolute reduction of 78 mg/dL (2.0 mmol/L). The total RNA dose administered correlated with the magnitude of LDL reduction (Pearson r = −0.68).

Durability

Follow-up extended to a maximum of 18 months; 15 of 35 participants had ≥1 year of follow-up. PCSK9 and LDL reductions at day 28 were consistent with the time-averaged values across follow-up, reflecting a stable response — the durability expected of a DNA-level edit persisting through hepatocyte turnover (hepatocyte lifespan ~200–300 days). gap/long-term-unknown — durability beyond ~12 months in humans, and across decades of hepatocyte regeneration, remains unproven.

Safety

Eventn (%)
Any adverse event26 (74)
Any grade ≥31 (3)
Any serious adverse event1 (3)
Any treatment-related AE11 (31)
Infusion-related reaction (all grade 1–2)7 (20)
Fatigue2 (6)
  • No dose-limiting toxic effects, no deaths, no withdrawals. All participants received the full planned dose.
  • Transient ALT elevations: 3 of 35 had ALT ≥2× ULN (peaks 2.0, 2.2, 2.4× ULN; 1 in the 0.7-mg/kg and 2 in the 1.0-mg/kg cohort), peaking on day 3–4 and falling below 2× ULN by day 8. Asymptomatic. No event exceeded grade 3.
  • One serious AE: grade 3 aspiration pneumonitis ~2 weeks post-infusion in a 0.45-mg/kg participant with gastroesophageal reflux disease and sliding hiatal hernia; hospitalized, discharged next day; assessed by the site investigator as unrelated to VERVE-102.
  • Pharmacokinetics: lipid-nanoparticle components cleared rapidly; mean terminal half-life of the primary LNP component was <20 h — consistent with transient base-editor exposure followed by a permanent genomic edit.

The authors emphasize that the GalNAc-LNP (a reformulation from the predecessor VERVE-101) was not associated with the infusion reactions, larger ALT rises, and thrombocytopenia reported with earlier in-vivo editing programs — supporting the field view that the lipid nanoparticle, not the editor cargo, is the principal driver of acute adverse events.

Extrapolation to the wiki’s core question

This is a human trial, so the usual model-organism extrapolation table mostly resolves to “yes”:

DimensionStatusNotes
Pathway conserved in humans?yesThe trial is in humans; the PCSK9–LDLR axis is the validated target
Phenotype conserved (LDL-lowering)?yesDemonstrated directly; magnitude (−62%) matches PCSK9 mAbs (40–60%)
Replicated / durable in humans?in-progressSingle Phase 1, n=35, interim; durability shown to ~12 mo; hard CV-outcome benefit predicted but untested

The geroprotective thesis (atherosclerosis is a cumulative-LDL-exposure disease, so a one-time durable lowering early in life should compound protection) is mechanistically supported but not directly tested here: the authors note that a 78-mg/dL reduction maintained over 20 years is predicted to cut ASCVD risk by >50% for most patients, citing prior Mendelian-randomization/consensus modeling — this is an extrapolation, not a measured endpoint. gap/needs-human-replication

Interpretive caveats

  1. Not powered for, and did not assess, cardiovascular outcomes. All efficacy here is biomarker (PCSK9, LDL-C). gap/needs-replication
  2. Non-prespecified interim analysis, single-arm, open-label, no formal statistical testing. Effect-size point estimates per small dose cohort (n=4–8) carry wide uncertainty; the 0.7-mg/kg LDL response (−33%) being numerically below the 0.6-mg/kg (−45%) cohort illustrates small-cohort noise.
  3. Short follow-up relative to the durability claim. “Durable” rests on ≤18-month data in a minority of participants; the cross-decade persistence that the cumulative-exposure rationale requires is unproven. gap/long-term-unknown
  4. Off-target / long-term safety not the focus of this interim report. Nonclinical off-target screening was done largely in cells from White donors; the cohort was 86% White. Genome-wide off-target surveillance in humans is the load-bearing future safety question for the platform — see in-vivo-base-editing. gap/long-term-unknown
  5. Selected, premedicated, inpatient-observed population with stable disease — safety may not transfer to broader, unmonitored outpatient use.

Hallmarks engaged

Limitations (authors’ stated)

Phase 1; selected population; premedication and inpatient observation; not powered for statistical effect size; never assessed cardiovascular disease; short, non-prespecified-interim follow-up precluding long-term safety (including off-target) and durability assessment; majority-White cohort and predominantly White-donor nonclinical off-target work. All participants are expected to enroll in a 15-year long-term follow-up study per regulatory guidance.

Conflict of interest

Sponsored by Verve Therapeutics, a wholly owned subsidiary of Eli Lilly. Many authors are Verve/Lilly employees; the study was designed by the sponsor and the first manuscript draft written by sponsor employees. Treat the framing accordingly; the quantitative tables are the primary evidence.

Cross-references

Footnotes

Graph View

Backlinks