Introduction
In vivo gene editing refers to delivering genome-editing machinery directly into a patient so that the edit is made inside target tissues, rather than manipulating cells outside the body and re-infusing them. The clearest clinical proof-of-concept for this in 2025–2026 comes from programs that deliver base editors systemically as a single intravenous infusion using lipid nanoparticles (LNPs). VERVE-101, for example, uses messenger RNA encoding an adenine base editor plus a guide RNA targeting PCSK9, packaged in an LNP and administered as a one-time intravenous infusion[1]. Similarly, YOLT-101 is an investigational in vivo therapy using adenine base editing delivered via GalNAc-modified LNPs to inactivate PCSK9 after a single intravenous dose[2].
The therapeutic promise of these “one-and-done” designs is not that they immediately prove reductions in myocardial infarction or mortality, but that a permanent edit could, in principle, replace lifelong adherence to daily or periodic lipid-lowering regimens—if durability, safety, and real-world feasibility hold up in larger and longer studies. That durability-and-safety question is precisely what early clinical trials are now testing, alongside mechanistic evidence for how LNPs, hepatocyte targeting, and base editing work in humans and translational models[3].
The Mechanism
Base editing is often described as “precision editing” because it can directly change one DNA base to another without requiring a double-strand DNA break (DSB). In a cardiovascular context, this matters because DSB-based nuclease editing can yield a spectrum of repair outcomes, while base editors and prime editors can “directly modify DNA sequences without inducing double-strand DNA breaks,” which is described as reducing the risk of some unintended outcomes such as “uncontrolled indels or large deletions”[4]. Consistent with this framing, base editors and prime editors have been reported to generate large deletions at approximately 20-fold lower frequency than Cas9 nucleases in the analysis highlighted by a 2025 Nature Biomedical Engineering editorial[4].
Mechanistically, the base editing approach emphasized in the 2025–2026 clinical programs is adenine base editing. In YOLT-101’s description, the base-editing complex catalyzes deamination of adenine (A) to inosine (I), which cells interpret as guanine (G)[2]. This yields a precise A-to-G substitution that disrupts normal PCSK9 mRNA splicing and introduces a frameshift mutation that inactivates PCSK9[2]. In VERVE-101, the intended A/T to G/C edit disrupts the PCSK9 splice donor site, inactivating PCSK9 in the liver[1].
Delivery is the other half of the mechanism. LNPs are described as “the most established platforms for the delivery of macromolecules including DNA, mRNA, and protein into cells,” with use dating to the 1990s and serving as the vehicle for the first FDA-approved RNAi therapeutic in 2018[5]. A key enabling concept is that ionizable cationic lipids complexed with cargo enter cells through endocytosis and become positively charged upon endosomal acidification, disrupting the endosome membrane and releasing cargo into the cytoplasm[5]. In practical terms, this enables transient intracellular expression of a base editor from mRNA (for example, VERVE-101’s ABE mRNA payload)[1], and transient/controlled exposure is discussed as one reason non-viral approaches such as LNPs and RNP delivery are being intensively explored for safety and off-target control[4].
PCSK9 and the cardiovascular use case
The liver has become the first—and still dominant—organ for in vivo editing because it is comparatively tractable for systemic delivery. As summarized in a 2023 review by Newby and Liu, “due to the availability of multiple efficient liver delivery methods, the first and most efficient in vivo editing demonstrations have targeted diseases that can be treated by editing hepatocytes”[5]. The PCSK9 target fits this paradigm: editing hepatocytes changes circulating PCSK9 protein levels and thereby modulates LDL receptor (LDLR) biology in the liver.
Both YOLT-101 and related “one-time infusion” concepts emphasize hepatocyte targeting using GalNAc ligands that bind the asialoglycoprotein receptor (ASGPR). YOLT-101’s carrier system is explicitly described as a GalNAc-modified LNP designed for enhanced delivery to hepatocytes[2], and the paper states that GalNAc directs LNPs to hepatocytes by targeting ASGPR, enhancing delivery via an LDLR-independent pathway[2]. A complementary mechanistic summary in a lipoprotein-disorder gene therapy review describes a GalNAc-conjugated LNP approach for PCSK9 editing that leverages hepatocyte uptake via ASGPR or LDLR-mediated endocytosis[6].
Once PCSK9 expression is reduced, the mechanistic intent is to enhance LDLR recycling. YOLT-101’s report explicitly links reduced PCSK9 expression to “enhanced recycling of the LDLR”[2]. The clinical hypothesis is that this could translate into durable LDL-C lowering after a single infusion, but it remains essential to distinguish biomarker reductions (PCSK9 and LDL-C) from unproven effects on cardiovascular outcomes endpoints in current early-phase trials[3].
Clinical evidence in 2025–2026
The most decision-relevant evidence for “one-and-done” cardiovascular base editing, as of May 2026, comes from two early clinical datasets: (i) VERVE-101 in the ongoing Heart-1 phase 1b trial and (ii) interim phase 1 results for YOLT-101 reported in Nature Medicine.
VERVE-101 in Heart-1
Heart-1 is described as an “ongoing, open-label, ascending dose, phase 1b study designed to assess the safety and tolerability of VERVE-101” in familial hypercholesterolaemia[3]. In an interim report, 10 patients with established atherosclerotic cardiovascular disease (ASCVD) were described as being enrolled, and all were characterized as high risk for cardiovascular events[3]. Despite oral lipid-lowering therapy, mean entry LDL cholesterol was reported as 193 mg/dL, and VERVE-101 was administered as a single peripheral intravenous infusion across four escalating dose cohorts (0.1, 0.3, 0.45, and 0.6 mg/kg) after pre-medication with dexamethasone and antihistamines[3].
Efficacy signals were reported as biomarker changes at day 28 and beyond. At 28 days, blood PCSK9 was reduced by 59% and 84% in patients treated with the 0.45 mg/kg dose, and by 47% in the patient receiving 0.6 mg/kg[3]. LDL-cholesterol decreased by 39% and 48% with the 0.45 mg/kg dose and by 55% with the 0.6 mg/kg dose[3]. The 55% LDL reduction was reported to persist for 6 months[3]. Separately, the same source describes that in a preclinical monkey study, LDL-cholesterol reduction lasted 2.5 years after a single dose[3].
Safety signals in this interim discussion included brief flu-like symptoms (including fever, headaches, and body aches)[3] and a temporary increase in liver enzymes that returned to normal within days[3]. Two cardiovascular events were reported during the study—one fatal cardiac arrest 5 weeks after treatment and one acute myocardial infarction a day after infusion[3] —with the independent safety board concluding that the events were probably related to the patients’ underlying disease and “not necessarily” to treatment, recommending continuation of enrollment without protocol changes[3].
YOLT-101 phase 1 interim in Nature Medicine
YOLT-101 is described as an investigational in vivo gene therapy using adenine base editing delivered via GalNAc-modified LNPs to inactivate PCSK9 and achieve sustained LDL-C reduction[2]. The interim report describes an ongoing clinical trial evaluating primary safety/tolerability and secondary outcomes (lowering PCSK9 and LDL-C) after a single intravenous dose in adults with heterozygous familial hypercholesterolemia (HeFH) and uncontrolled LDL-C. Six participants (three men and three women) received escalating doses of 0.2, 0.4, or 0.6 mg/kg, and no grade ≥3 adverse events occurred[2].
The most common adverse events were reported as “transient and self-limited infusion-related reactions and elevations in liver enzymes”[2]. At 24 weeks in the 0.6 mg/kg cohort (n = 3), reductions were described as durable, with sustained reductions of 74.4% in circulating PCSK9 and 52.3% in LDL-C[2].
The paper also provides an explicit off-target evaluation frame in primary human hepatocytes, describing net A-to-G editing at on-target and 62 candidate off-target sites across three donors[2], with a stated next-generation sequencing (NGS) detection limit of 0.1% (values below this threshold indicated)[2]. For RNA-level off-target concerns, it reports that, following SNP-based analysis, no significant additional A-to-I RNA edits were detected at the EC90 dose compared to untreated controls (P-value = 0.1385, one-sided Wilcoxon-Mann-Whitney test)[2].
A snapshot comparison
The table below summarizes the most concrete, quoted clinical efficacy and safety details available from the provided sources.
Beyond the liver
Even as the liver remains the most accessible organ for systemic nucleic-acid delivery, multiple lines of work are testing whether LNP composition and cargo format can push editing into other tissues with useful efficiencies. A 2024 study delivering a stable CRISPR–Cas9 ribonucleoprotein (RNP) via tissue-selective LNPs reports genome-editing levels of 16–37% in the liver and lungs of reporter mice after single intravenous injections of iGeoCas9 RNP–LNPs[7]. In a more specific readout, imaging and flow quantification showed an average of 37% editing in the liver with one LNP formulation (FX12m) and an average of 16% editing in the lungs with another (FC8m), in n = 5 mice[7].
Crucially, the same study shows that such organ-selective formulations can be extended from reporter assays to therapeutically relevant genes. Using NGS, the authors report successful editing of PCSK9 in mouse liver with an average of 31% editing, and editing of the lung disease gene Cftr with an average of 19% editing in the lungs using liver- and lung-favoring formulations, respectively[7]. Tissue collection for these measurements is described as occurring 10 days after injection in wild-type mice under similar experimental procedures[7].
These data do not yet establish clinical feasibility for lung-directed cardiovascular prevention, but they do show that “beyond-liver” biodistribution can be engineered and quantified in vivo, and that lung editing is not merely theoretical when RNP payloads and LNP chemistries are co-optimized[7].
Open questions and limitations
The 2025–2026 clinical results should be read as early biomarker evidence, not as outcome-proven cardiovascular prevention. The Heart-1 interim perspective explicitly notes unanswered questions about “longterm efficacy,” and emphasizes that the key unknown is not only cholesterol levels but “hard clinical endpoints”[3]. Both Heart-1 and YOLT-101 are small (10 patients and 6 participants reported, respectively), which limits inference about rare adverse events and about heterogeneity across real-world populations[2, 3].
Safety and off-target editing remain central uncertainties even for DSB-free editors. A dedicated base/prime editing review states that off-target base editing can occur as guide-independent spurious RNA editing or genomic DNA editing, and also as guide-dependent off-target editing at mismatched sites engaged by the RNP[5]. The same review highlights a practical limitation of base editors: careful positioning is required to put the target within the optimal editing window while excluding undesired bystander edits[5]. In preclinical LNP-mediated nuclease editing of ANGPTL3, one group reports no evidence of editing at any of the nine top-predicted off-target sites they interrogated[8], while in a separate dual-AAV cytosine base editing system, the authors report that AncBE4max induced “low but significant gRNA-independent editing” along an induced R-loop in an orthogonal assay[9] —illustrating that “off-target” has multiple mechanistic forms that must be evaluated with appropriate assays.
Delivery strategy also shapes both efficacy and safety. GalNAc-targeting can “rescue” hepatocyte delivery even when LDLR-mediated uptake is impaired. In LDLR knockout non-human primates (a severe LDLR-deficiency model), standard LNPs at 2 mg/kg produced minimal target-site editing and little reduction in blood ANGPTL3 protein[10], whereas GalNAc-LNPs at the same 2 mg/kg dose achieved 60% whole-liver ANGPTL3 editing and >90% lowering of blood ANGPTL3 protein, alongside ~35% lowering of blood LDL-C and ~55% lowering of triglycerides (data not shown for triglycerides)[10]. This is encouraging for delivery robustness, but it also underscores that formulation changes can be the difference between “minimal” and substantial editing—making manufacturing control and reproducibility a practical barrier for scale-up and broad access[10].
Finally, even when transient LNP delivery is framed as advantageous for safety due to controlled expression, cost and implementation remain open questions for a therapy that is administered once but must justify lifelong risk/benefit tradeoffs. The clinical data to date document infusion reactions, liver enzyme elevations, and—in Heart-1—cardiovascular events in a very high-risk population, reinforcing the need for careful trial design, longer follow-up, and transparent adjudication as programs move beyond proof-of-concept biomarker endpoints[2, 3].
Bottom line
As of May 2026, the strongest evidence that “one-and-done” cardiovascular gene editing is technically feasible in humans comes from small, early-phase trials of in vivo adenine base editing targeting PCSK9. Heart-1 interim data report LDL-C reductions up to 55% with persistence for 6 months in the highest-dose cohort described, with preclinical durability in monkeys reported out to 2.5 years after a single dose[3]. YOLT-101 interim data report sustained reductions at 24 weeks of 74.4% for PCSK9 and 52.3% for LDL-C in the 0.6 mg/kg cohort (n = 3), with no grade ≥3 adverse events reported[2]. The science is moving quickly, but the clinically decisive questions—rare toxicities, long-term off-target risks, redosing strategy, and cardiovascular outcomes—remain open and are explicitly recognized as the next hurdles for the field[3, 5].
