Thirteen lines, eleven of them therapeutic, including the three that are closed.
The dead ones are listed with the measurement that closed them. A page showing only the live options would be advocacy rather than a record. A longer list is not better news: most of these lines are closed, conditional, or waiting on the same measurement as everything else. Current care is listed last, unranked, so the comparison sits on the page rather than being assumed.
None of this is a treatment, or close to one.
The strongest statement available is that one route now has a designed guide, a genome wide safety scan, and a costed validation experiment. That is a long way from a therapy, and saying so is what makes the rest believable.
Leading, and still unvalidated
A complete in silico design using an ABE8e-SpRY editor targeting c.311G>A. A genome-wide off-target scan identified 22 missense and 26 protein-changing candidate off-target sites, with zero sites within two mismatches. What blocks it is that nobody has run it, which requires a wet-lab cellular bench.
Conditional, meaning something specific is missing
The better designed option at this site and the worse delivered one: published in vivo correction in heart remains roughly 1 to 6 percent, far below the 60 percent therapeutic threshold. Base editing has the efficiency, prime editing has the sequence flexibility, and neither has reached a human heart cell.
AAV9-MOG1 gene therapy is advancing in preclinical mouse models. However, computational screening of small molecule chaperones like agmatine confirmed no viable binding pocket in the N-terminal domain. Furthermore, if the mutant channel actively interferes at the cell surface, escorting more of it out worsens the current.
The 6,048-base coding sequence exceeds single-AAV capacity. Dual-AAV intein trans-splicing rescues current in mice, but the extreme viral vector doses and cardiac mosaicism (leaving roughly a quarter of myocytes untreated) create dangerous conduction heterogeneity.
Delivering a compact bacterial sodium channel or nerve channel avoids the mechanism question because it does not associate with human Nav1.5. However, expressing foreign bacterial proteins in human heart tissue poses lifetime cardiac immunogenicity, and published computational rescue relied on accelerated inactivation rather than current loss.
Inhibiting developmental pathways (such as Wnt signalling) quadrupled sodium current in heart cells derived from Brugada patients. However, the targeted pathways operate body-wide rather than specifically in the heart, creating off-target organ risks.
Weak
ADAR-mediated adenosine-to-inosine editing exhibits high transcriptome specificity, but adult human heart ranks 53rd of 54 tissues in ADAR expression, and cardiac editing substrate abundance is 0.0045 percent.
Telling the broken copy apart from the working one requires discriminating a single letter (c.311G>A) in a non-selective sequence window, risking collateral knockdown of the healthy copy.
Dead, with the measurement that closed it
A closed route is a result. It is one fewer thing for the next person to try.
No resolved alpha-subunit interface structure exists, and an interaction blocker cannot replace the missing current. The route is closed unless physical interaction at the surface is experimentally proven.
The most attractive route on this list, because it did not depend on resolving the mechanism first. It needed a reserve of discarded message to redirect. Across 827 human hearts, that reserve measured 0.0045 percent, roughly 300 times smaller than in brain SCN1A. The theoretical ceiling is a 1.06-fold increase (taking 31.3 percent to 33.1 percent), against 1.5-fold needed.
Premise refuted by this project's own calibration test on 25 July 2026. The predictor it rested on caught one of four variants already known to break this channel, and helping an interfering mutant reach the surface could worsen arrhythmias.
Current care: an implanted defibrillator and existing drugs. For most people carrying a variant like this one it does not apply, because guideline indications for a device require a prior cardiac arrest, a documented sustained arrhythmia, or fainting from an arrhythmia together with a spontaneous type 1 pattern. A device also does not treat the disease: it stops an arrhythmia after it has already started, and does nothing to the sodium current, the protein or the gene. For most of the people a therapy would be developed for, the comparator is no treatment at all.
What changed between August and September 2026
A month of computational simulation, off-target refinement, and mathematical bounding resolved several assumptions that had governed these rankings:
A 1.41-million-atom molecular dynamics production run verified that N-terminal GFP fusion does not collapse the DEKA selectivity filter (5.24 angstroms radius, continuous 12.3-water hydration column, relaxed POPC bilayer), falsifying pore collapse artifacts and validating GFP fusion for trafficking assays.
A structure-aware off-target screen identified 22 missense and 26 protein-changing candidate off-target sites across the human genome, with zero sites within two mismatches. Protocol 3 was established with a binding week 12 stopping rule.
Computational docking against the Nav1.5 N-terminal domain confirmed the absence of a viable pocket for small molecule chaperones like agmatine, confirming that chaperone approaches must rely on protein escorts like MOG1.
Proved mathematically that whole-cell current and allele-resolved surface abundance together bound dominant-negative impairment without requiring physical separation of single-channel currents (I/I0 < Sw/S0).
Substituting Endoproteinase Lys-C for trypsin preserves position 104 intact on both alleles, yielding matched 26-residue peptides differing by 28.04 Daltons for quantitative mass spectrometry.
Cable modeling showed conduction velocity scales with the square root of current down to 50 percent, but drops steeply below 30 percent, explaining why the R104Q 31.3 percent boundary triggers conduction block and fatal arrhythmias.