Gene Therapy in Arrhythmia
Contents. #Introduction · #Definition · #Key Concepts · #Gene Replacement · #Gene Silencing · #Suppression-Replacement · #Gene Editing · #Governance and Risk · #How Much Correction Is Enough · #State of Play by Disease · #Safety · #Rationale and Mechanisms · #Contradictions / Open Questions · #Connections · #Sources
Detailed pages. concepts/AAV-Gene-Delivery · concepts/Gene-Silencing-Therapy · concepts/SupRep-Therapy · concepts/CRISPR-Cas9-in-Channelopathies · concepts/Gene-Editing-Risk-Benefit-Framework · concepts/Gene-Therapy-Transduction-Threshold
Introduction
Gene therapy for inherited arrhythmia is the attempt to treat the molecular cause of a monogenic arrhythmia rather than its electrical consequence — replacing a missing protein, silencing a mutant transcript, or correcting the variant itself. (sources/gene-therapy-arrhythmia-cjca-2026, rating: high)
- The target population is defined by what conventional therapy cannot do. Beta-blockers, flecainide, quinidine, sympathetic denervation, catheter ablation and the ICD treat risk and symptoms but not mechanism; the ICD in a young person carries a lifetime of lead complications, inappropriate shocks and psychological cost, and exercise restriction in PKP2 disease removes something patients value. (sources/gene-therapy-arrhythmia-cjca-2026)
- Five programmes are in human trials as of 2026 — three for PKP2-ARVC, one for RYR2-CPVT1, one for PLN p.R14del cardiomyopathy. Three years ago the count was one.
- Cargo capacity, not biology, decides the strategy. AAV carries ~4.5 kb, which excludes RYR2 and SCN5A outright and forces pathway-targeted, silencing or editing approaches for the largest and most important genes. (sources/gene-therapy-arrhythmia-2025, rating: high)
- Partial correction is enough — for arrhythmia. Roughly 30–50% of cardiomyocytes must be corrected to abolish triggered activity, against >70% for structural repair. This is the single most encouraging fact in the field and the reason channelopathies are the beachhead. (sources/gene-therapy-arrhythmia-cjca-2026)
- Durability is a platform problem, not a disease problem. AAV is one-shot and blocked from re-dosing by the immunity it induces; ASOs and siRNAs are re-dosable but need repeat administration. Which is preferable depends on whether the therapeutic window is wide or narrow.
- Safety is now the binding constraint. Four deaths in 2025 across AAV programmes have moved dose selection, capsid engineering and immunomodulation to the centre of the field.
Abbreviations. AAV, adeno-associated virus; ACM, arrhythmogenic cardiomyopathy; APD, action potential duration; ARVC, arrhythmogenic right ventricular cardiomyopathy; ASO, antisense oligonucleotide; BrS, Brugada syndrome; CPVT, catecholaminergic polymorphic ventricular tachycardia; CRDS, calcium release deficiency syndrome; CRISPRi, CRISPR interference; DAD, delayed afterdepolarisation; EAD, early afterdepolarisation; ICD, implantable cardioverter-defibrillator; LQTS, long QT syndrome; shIMM, shRNA-immune cDNA; shRNA, short hairpin RNA; siRNA, small interfering RNA; SQTS, short QT syndrome; Sup-Rep, suppression-replacement; vg/kg, vector genomes per kilogram.
Definition
Gene therapy in arrhythmia comprises nucleic-acid therapeutics that alter gene expression or sequence in cardiomyocytes in order to prevent arrhythmia. Three modalities are in use:
| Modality | What it does | Best suited to |
|---|---|---|
| Gene replacement | Delivers a functional cDNA or mRNA | Haploinsufficiency — PKP2, CASQ2, TMEM43 |
| Gene silencing | Reduces a transcript via siRNA, shRNA, ASO or CRISPRi | Gain-of-function and dominant-negative variants — RYR2, CALM, PLN, KCNH2 |
| Gene editing | Corrects the DNA sequence via base, prime or nuclease editing | Recurrent point mutations — SCN5A p.T1307M, RBM20 p.R636Q |
Suppression-replacement is a hybrid of the first two and is treated separately below.
What is excluded. Cell therapy, cardiac regeneration, and ex vivo engineering of cells for delivery are separate fields and are not covered here. Nor is gene therapy for acquired structural disease counted as arrhythmia gene therapy except where the endpoint is arrhythmic — post-infarct VT and atrial fibrillation are included as future directions because that is the endpoint being measured.
Key Concepts
- Monogenic diseases are the ideal substrate — which is why this field exists at all. One gene, one mechanism, one protein to restore. The same property that makes these diseases devastating in the young makes them tractable.
- The 4.5 kb cargo ceiling is the organising constraint of the whole field. It determines which genes can be replaced (CASQ2, PKP2, TMEM43 — yes; RYR2, SCN5A — no), which must be silenced instead, and why so much effort goes into small workaround proteins such as MOG1 and SCN10A-short. (sources/gene-therapy-arrhythmia-cjca-2026, rating: high)
- One-shot versus re-dosable is a genuine therapeutic choice, not a manufacturing accident. A single AAV dose suits a disease with a wide therapeutic window. Where over-correction is itself dangerous — silencing RYR2 toward CRDS, silencing KCNH2 toward long QT — a titratable, stoppable agent is safer even though it must be given repeatedly.
- Partial correction is sufficient for electrophysiological endpoints because corrected cells suppress uncorrected neighbours through electrotonic coupling. See concepts/Gene-Therapy-Transduction-Threshold. (sources/gene-therapy-arrhythmia-cjca-2026)
- Pre-existing anti-AAV immunity closes the door on re-dosing. Most adults carry neutralising antibodies to common serotypes, which is why trials exclude seropositive patients rather than trying to overcome their immunity — and why paediatric patients, whose episomes dilute as the heart grows, have no second chance.
Gene Replacement
Full detail: concepts/AAV-Gene-Delivery.
Adding a functional copy of a gene, delivered as cDNA or mRNA. It works when disease arises from haploinsufficiency — one working allele is not enough, and adding one more restores function.
- PKP2-ARVC is the paradigm: most variants are truncating and subject to nonsense-mediated decay, so replacing the gene addresses the mechanism exactly. It is why PKP2 reached clinical trials first. (sources/gene-therapy-arrhythmia-cjca-2026, rating: high)
- CASQ2 replacement rescues both CPVT2 and CPVT1 — the latter because overexpressing CASQ2 increases RYR2 refractoriness, sidestepping the impossibility of packaging RYR2 itself.
- TMEM43 is the instructive counter-example: truncating variants occur in healthy people, implying haploinsufficiency is tolerated, while the p.S358L missense is malignant. Replacement still worked in the mouse model, which means the mechanism there is not simple loss of protein.
- Delivery is by intravenous AAV, with the liver taking most of the dose. Cardiac-selective capsids and local delivery routes are under development but none has been shown superior in humans.
Gene Silencing
Full detail: concepts/Gene-Silencing-Therapy.
Reducing expression of a transcript without touching the DNA — by siRNA, shRNA, ASO or CRISPR interference. It is the right tool for gain-of-function and dominant-negative disease, where the mutant allele actively harms and adding wild-type protein does not help.
- Allele-specific silencing is elegant but does not scale. Constructs must be designed per variant, and no gene in this field has few enough variants to make that practical.
- Fatty-acid conjugation solved the delivery problem that kept cardiac ASOs in the laboratory — palmitate and docosanoic acid conjugation give muscle tropism, and both the CALM1 and PLN ASOs depend on it. (sources/gene-therapy-arrhythmia-cjca-2026)
- The first cardiac siRNA trial is running. PULSE (NCT07241104) delivers a subcutaneous siRNA against the PLN p.R14del transcript — the only non-viral, re-dosable programme among the five current trials.
Suppression-Replacement
Full detail: concepts/SupRep-Therapy.
A single construct carries an shRNA that knocks down the endogenous gene and an interference-immune (shIMM) cDNA that restores the protein. Both endogenous alleles are suppressed; a normal copy is supplied.
- It is the only mutation-agnostic strategy in the field. One construct covers every variant in a gene, and the calmodulinopathy version covers an entire three-gene family with one therapeutic.
- The stoichiometry is the therapy. Too much suppression without replacement worsens the phenotype being treated; too much replacement without suppression produces the opposite disease. No treated animal has yet had the shIMM-to-endogenous ratio measured. (sources/gene-therapy-arrhythmia-cjca-2026, rating: high)
Gene Editing
Full detail: concepts/CRISPR-Cas9-in-Channelopathies.
Correcting the sequence itself — nuclease editing, base editing, or prime editing. Editing is attractive where a single recurrent variant accounts for most disease, and where permanent correction is worth the immunogenicity and off-target risk of a dual-vector Cas construct.
- Base editing has an unexpectedly favourable dose–response. In SCN5A p.T1307M mice, 20% editing prevented carbachol-induced VT without normalising the QT interval — the clearest demonstration that a partial biochemical correction can produce a complete antiarrhythmic one.
- The size problem reappears. Standard Cas variants and base editors usually exceed AAV capacity and must be split across two vectors with split-intein reconstitution.
Governance and Risk
Full detail: concepts/Gene-Editing-Risk-Benefit-Framework.
Germline editing is prohibited; somatic editing of the heart is permissible under research frameworks but carries off-target, genotoxicity and immunogenicity risks that no cardiac programme has yet had to characterise in humans.
- The risk calculus is unusual here. These are often young, otherwise healthy people with an excellent device-based alternative. The tolerance for a serious adverse event is therefore far lower than in, say, Duchenne muscular dystrophy — a point the preclinical literature is not designed to address. (sources/gene-therapy-arrhythmia-cjca-2026, rating: high)
- Consent is complicated by the promise. "One dose instead of a lifetime of tablets and an ICD" is a compelling offer to a 20-year-old, which is exactly why the trials are safety-primary and ascending-dose.
How Much Correction Is Enough
Full detail: concepts/Gene-Therapy-Transduction-Threshold.
No vector transduces every cardiomyocyte, so the field's central question is the minimum fraction that must be corrected.
- For arrhythmia, ~30–50%. Source–sink buffering means only 70–80 contiguous cells need to depolarise together to fire in 1-D, so corrected cells can actively hold their neighbours in line. Animal data agree: CPVT 40–66%, CASQ2 p.D307H >33%, LQT3 20%, LQT1 30%, SQT1 47%.
- For structural repair, >70% — and in cardiomyopathy, fibrosis and reduced gap junction coupling raise the arrhythmic threshold toward the structural one. The two numbers are not contradictory; they measure different things.
State of Play by Disease
| Disease | Gene(s) | Modality | Furthest stage | What is established |
|---|---|---|---|---|
| CPVT1 | RYR2 (~70%) | Silencing; pathway modulation | Phase Ib (NCT07148089) | Allele-specific silencing and CRISPR both abolish arrhythmia in mice; CaMKII inhibition and CASQ2 overexpression work across genotypes; the trial delivers AAV8-CASQ2 |
| CPVT2 | CASQ2 (~5%) | Replacement | Preclinical | AAV9-CASQ2 restores junctional SR ultrastructure and protects to 12 months; >33% transduction needed to abolish arrhythmia, less prevents sustained VT |
| LQT1 / LQT2 | KCNQ1 (~50%), KCNH2 (~35%) | Sup-Rep | Preclinical (rabbit, hiPSC) | One construct covers four LQT1 variants; QT and dispersion normalised in rabbits on ~30% mutant knockdown |
| LQT3 | SCN5A (5–10%) | Base editing | Preclinical | 20% editing prevents VT without QT normalisation; dual-AAV ABE8e-SpRY achieved 54% efficiency |
| SQT1 | KCNH2 (gain) | Sup-Rep, silencing | Preclinical (rabbit) | 60% mutant silencing recovers QTc and apicobasal APD₉₀ heterogeneity; IKr normalised in 47% of cells |
| Calmodulinopathy | CALM1/2/3 | CRISPRi, Sup-Rep, ASO | Preclinical (hiPSC, mouse) | A universal three-shRNA Sup-Rep construct corrects all three genotypes; ASO needs >85% silencing and lasts 4 weeks |
| Brugada | SCN5A (20%) | Pathway modulation | Preclinical | MOG1 corrects trafficking; SCN10A-short raises INa directly; no trial, and current therapy is already effective |
| PKP2-ARVC | PKP2 (~40% of ARVC) | Replacement | Three Phase I trials | Haploinsufficiency mechanism; late treatment stabilises function but does not abolish arrhythmia vulnerability |
| TMEM43-ACM | TMEM43 p.S358L | Replacement or allele-specific knockdown | Preclinical | AAV replacement improved survival and function on only 10–30% transduction, including after onset |
| PLN-CM | PLN p.R14del | Silencing | Phase I (NCT07241104) | ASO tripled mouse survival; CRISPR knockout and DWORF overexpression also effective |
| DSP / FLNC / RBM20 / LMNA | Various | Cx43 delivery, CRISPR activation, base editing, Sup-Rep | Early preclinical | Each has one proof-of-concept study, mostly in mouse or hiPSC-CM |
| Post-MI VT | None (acquired) | Refractoriness or conduction modulation | Preclinical | Dominant-negative KCNH2, Nav1.4, Cx32, SCN10A-short and SERCA2a all reduce inducible VT; none in trial |
| Atrial fibrillation | None (acquired) | Refractoriness or conduction modulation | Preclinical | KCNH2-G628S, Cx43, Cx40 and TASK-1 knockdown reduce AF burden in pigs; atrial-specific delivery is unsolved |
The pattern is clear: diseases where a small protein can be replaced or a single transcript silenced are furthest along; diseases where the culprit channel is too large to package, or where the arrhythmia is acquired, are still entirely preclinical. (sources/gene-therapy-arrhythmia-cjca-2026, rating: high)
Safety
- Four deaths in 2025 across AAV programmes. A Rocket Pharmaceuticals death in the pivotal phase 2 RP-A501 trial in Danon disease, potentially related to a newly added C3 inhibitor in the pre-treatment immunosuppressive regimen; and three FDA-disclosed deaths from acute liver failure after Sarepta AAVrh74 therapies for Duchenne and limb-girdle muscular dystrophy. All occurred at doses at or above the high-dose range (>1 × 10¹⁴ vg/kg). (sources/gene-therapy-arrhythmia-cjca-2026, rating: high)
- The arrhythmia trials dose below that ceiling but not comfortably below it. The PKP2 programmes run at 2.0–6.0 × 10¹³ vg/kg.
- Complement activation is the characterisable hazard. In the Danon programme, high-dose AAV9 produced grade 4 complement-mediated thrombotic microangiopathy requiring renal replacement therapy and transplantation, with complement activation more marked in adults than children; LVEF <40% became an exclusion criterion thereafter. (sources/aav9-danon-nejm-2025, rating: high)
- Immunosuppression is itself morbid. Three-drug regimens have caused glucocorticoid myopathy and opportunistic infection — a serious problem in a population that already has a cardiomyopathy. (sources/aav9-danon-nejm-2025)
- Paediatric patients carry a double penalty — episomal dilution as cardiomyocytes divide, and no re-dosing because the first dose induces immunity. The CPVT1 trial's decision to extend down to age 7 makes this an active question rather than a theoretical one.
Rationale and Mechanisms
The case for gene therapy rests on three claims, each of which the field has partly established and partly not.
Mechanism is addressable. In channelopathies the arrhythmia is a direct consequence of one protein's behaviour, and correcting that protein in a minority of cells suppresses the trigger across the tissue. In cardiomyopathy the protein is structural, correction must be per-cell, and fibrosis raises the threshold — so the same platform is more likely to work in RYR2 than in PKP2, which inverts the order in which the trials actually happened. (sources/gene-therapy-arrhythmia-cjca-2026, rating: high)
The comparator is weak. An ICD prevents death but not arrhythmia, does not prevent disease progression, and in the young carries a high lifetime burden of complications and shocks. Exercise restriction is a real cost in PKP2 disease. A one-time therapy that removed the substrate would be a different category of treatment, not an increment.
The window may be early. Late intervention in the Pkp2 model stabilised function but left arrhythmia vulnerability intact, and the modelling work says fibrosis and reduced coupling raise the number of cells that must be corrected. If that holds in humans, the current trial designs — adults who already have an ICD — may be testing the therapy where it is least likely to work. (sources/gene-therapy-arrhythmia-cjca-2026)
Where the approach cannot reach: polygenic disease such as most Brugada syndrome, acquired structural disease with established scar, and anything requiring correction of a gene too large to package and too heterogeneous to silence cheaply.
Contradictions / Open Questions
- The 30–50% transduction threshold rests on inferred, not measured, correction. Several preclinical studies inferred transduction from the degree of mutant transcript silencing rather than counting transduced cells, and in the LQT1 rabbit work the fraction of transduced cardiomyocytes was never determined. The number may be right; it has not been measured in the tissue it describes. (sources/gene-therapy-arrhythmia-cjca-2026, rating: high)
- Suppressing the trigger and avoiding reentry may be incompatible. Every favourable result concerns triggered activity; every warning concerns dispersion of repolarisation. No study has measured both endpoints in the same animal with histology to confirm vector distribution, so "safe at 30%" remains inference rather than observation.
- The PKP2 trials may be enrolling too late. All three require an existing ICD, which means an established phenotype and possibly an established substrate. The mouse data show that late treatment improves function without removing arrhythmia vulnerability — so a trial powered on arrhythmia burden could fail for reasons that say nothing about the therapy's biology.
- Silencing RYR2 risks converting CPVT into its opposite. RYR2 loss of function causes calcium release deficiency syndrome. Knockdown models have not shown CRDS-like remodelling, but organ-level susceptibility has never been tested, and the CPVT1 trial does not list CRDS screening among its endpoints.
- Whether PLN p.R14del is gain or loss of function decides which therapy is rational. The longstanding model was constitutive SERCA2a inhibition; more recent evidence suggests inhibitory capacity is impaired and disease arises from aberrant SR/ER cluster formation. Silencing is rational under the second model; SERCA2a activation is rational under the first. DWORF working without changing SERCA2a activity favours the second.
- One-shot AAV is a poor fit for diseases needing titration. In LQTS and SQTS the therapeutic window runs from insufficient correction to the opposite channelopathy, and a single non-adjustable dose cannot be walked back. The modality that is furthest advanced is not the modality best suited to the largest group of diseases.
- Variant-specific versus universal constructs remains unresolved in practice. Sup-Rep solves it in principle for any single gene; allele-specific editing solves it only for recurrent variants. Which wins will be decided by manufacturing cost and regulatory precedent as much as by biology.
- Paediatric durability has no solution. Episomal AAV dilutes as cardiomyocytes divide; immunity blocks re-dosing; integrase-based vectors carry oncogenic risk. The field is about to treat children anyway.
- Preclinical effect sizes are almost certainly inflated. In a systematic review of 26 AAV arrhythmia studies, not one reported failure. The OR of 0.06 for ventricular arrhythmia inducibility is not a credible estimate of human effect. (sources/aav-gene-therapy-arrhythmia-hr-2024, rating: medium)
Connections
- The modalities: concepts/AAV-Gene-Delivery · concepts/Gene-Silencing-Therapy · concepts/SupRep-Therapy · concepts/CRISPR-Cas9-in-Channelopathies
- How much correction is needed: concepts/Gene-Therapy-Transduction-Threshold — the field's central quantitative question
- Governance: concepts/Gene-Editing-Risk-Benefit-Framework — somatic editing limits and off-target risk
- The disease templates: entities/CPVT · entities/PKP2 · entities/PLN — the three with human trials
- The channelopathies targeted: entities/Long-QT-Syndrome · entities/Short-QT-Syndrome · entities/Brugada-Syndrome · entities/SCN5A · concepts/Calmodulinopathy
- The cardiomyopathies targeted: entities/ARVC · entities/TMEM43 · entities/CASQ2 · entities/RYR2 · concepts/Arrhythmogenic-Cardiomyopathy
- Where the safety experience comes from: concepts/Danon-Disease · concepts/Pompe-Disease — the two cardiac AAV programmes with human follow-up
- What it competes with: entities/ICD · concepts/Left-Cardiac-Sympathetic-Denervation · entities/Amiodarone — devices and drugs that manage risk without modifying mechanism
- The acquired-disease frontier: entities/Atrial-Fibrillation · concepts/VT-Ablation-Ischemic-Cardiomyopathy
- The sources: sources/gene-therapy-arrhythmia-cjca-2026 · sources/gene-therapy-arrhythmia-2025
Sources
Reviews — arrhythmia gene therapy
Systematic review
Editing, delivery and governance
Human cardiac AAV experience outside arrhythmia
Disease-specific