Therapeutic routes
Nine routes, ranked, including the two that are closed.
Two of these are dead and one does not apply. They are listed anyway, each with the measurement that closed it. A page showing only the live options would be advocacy rather than a record.
What these are being compared against
There is a common assumption that a therapy for this condition would have to beat an implanted defibrillator. For most people carrying a variant like this one, that comparison does not apply. 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 on an electrocardiogram. Most carriers are asymptomatic, meet none of those, and therefore have no protection at all.
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. It carries real measured harm as well: inappropriate shocks, meaning a full discharge delivered to a conscious person who did not need one, occurred in 20 percent of patients at a crude rate of 4.7 percent per year, across 63 studies and 4,916 patients with inherited rhythm conditions. Olde Nordkamp et al., Heart Rhythm 2016, PMID 26385533
So for most of the people a therapy would be developed for, the comparator is no treatment at all. That is the correct framing and it is the one used throughout this site.
The nine routes
1. DNA base editing
leadingWhat blocks itDelivery to human heart muscle at the efficiency seen in mice is unproven, and nothing has been tested in a cell.
An adenine base editor changes a single DNA letter without cutting the strand. It leads because this exact gene has already been corrected in a living mouse heart, at up to 99.2 percent of transcripts from one injection, with the disease signature disappearing above 60 percent correction. The design here is complete: one guide, the target letter sitting squarely in the editing window, and no neighbouring letter the editor could change by mistake.
2. Prime editing
conditionalWhat blocks itPublished cardiac efficiency is roughly nine times below the base-editing precedent, and far below the 60 percent that mattered.
Prime editing writes the corrected letter from an RNA template rather than chemically converting it. At this site it is the better-designed option: it recovers a standard docking sequence that base editing cannot use at all, and it has to satisfy three separate requirements at every wrong site rather than one, which removes the overwhelming majority of them. It loses on delivery, not on design.
3. RNA editing
weakWhat blocks itHeart muscle carries almost the least of the enzyme the whole approach borrows, ranking 53rd of 54 human tissues.
This edits the message copied from the gene rather than the gene itself, so it is reversible and never alters the genome. The chemistry is an exact match: the repair needed is A to G, and the enzyme already present in every cell does precisely that conversion. The specificity result is strong, with the 30-letter designs hitting exactly one site across the entire transcriptome. The problem is the target organ.
4. Interaction drug
conditionalWhat blocks itNo target is defined and no assay exists. It waits entirely on the mechanism experiment.
If the broken copy interferes with the working one at the cell surface, then the thing to target is the interaction itself rather than the channel. There is no published structure of that interface to build on, so this branch starts closer to zero than it feels.
5. Raising output from the healthy copy
deadWhat blocks itThe reserve it would redirect measures 0.0045 percent, against roughly the 1.5-fold increase the route needed.
The idea was to make the working copy produce more channel, and it was attractive because it was the only route that did not depend on resolving the mechanism first. It needs a pool of non-productive message to redirect. Across 827 human heart samples that pool is about 300 times smaller than the equivalent pool in brain, where the same strategy is already proven in children. The ceiling is a 1.06-fold increase, taking 31.3 percent to 33.1 percent.
6. Silencing the broken copy
weakWhat blocks itThis change is the hardest possible class to tell apart from the healthy copy.
Shutting off only the broken copy means distinguishing two sequences differing by one letter, and this particular substitution is the most conservative one available.
7. Folding corrector
deadWhat blocks itRefuted by this project's own calibration test.
A small molecule was to hold the protein steady while it folded. Given four variants already known to break this channel, the folding-stability method underneath the idea identified one. The route also carries a hazard: if the broken copy interferes with the working one, helping more of it reach the surface could make things worse.
8. Gene replacement
not applicableWhat blocks itThe gene is 6,048 letters of coding sequence against roughly 4,700 that the standard delivery vehicle carries.
Delivering a whole working copy is standard practice for some conditions. This gene does not fit in the vector. That is a field-level problem, not one this project can move.
Implanted defibrillator, and existing drugs
current careWhat blocks itNeither treats the disease. A device stops an arrhythmia after it starts.
This is what exists today, and for people who qualify a device is what keeps them alive. It does not change the sodium current, the protein, or the gene, and most carriers do not qualify for one at all.
The route that died on a measurement
Raising output from the healthy copy was the most attractive route on this list, because it was the only one that did not depend on resolving the mechanism first. It needed a reserve of non-productive message to redirect, and the analysis said in advance that the route would die if that reserve turned out to be a few percent.
It was attractive because it was mechanism-independent. It is dead because the substrate is not there. Those are independent reasons, and a measurement rather than an assumption settled it.
Why nobody has done this already
The usual answer is that a condition is too rare to attract attention. That is not the reason here, and the real one is more interesting because it will not change.
The mouse study that makes gene editing look ready for the heart corrected a different variant in this same gene, Scn5a T1307M, which causes long QT syndrome type 3. That is a gain of function change, the opposite of this one. Its effect shows up as a measurable interval on a mouse electrocardiogram, so a researcher can inject, record, and watch the number move. A single injection corrected up to 99.2 percent of transcripts, and above 60 percent correction the disease signature disappeared. Qi et al., Circulation 2024, PMID 37965733
Brugada syndrome has no equivalent readout. Its signature depends on a difference in electrical recovery across the thickness of the right ventricular wall, and mouse hearts do not reproduce that gradient. So an editing study in this condition can measure current in isolated cells, and it can show the letter was corrected, but it cannot point at an animal and say the disease went away.
That is the difference between one paper and five years of work. No amount of funding removes it.
The collision that two separate analyses found
The gene this variant sits in has a close relative, and the region around position 104 is similar enough between them that molecules designed against one can reach the other. Two independent analyses found this separately: one working on an RNA molecule, one on a DNA one, sharing no code and no scoring method.
That makes it a property of the target sequence rather than of any one design. Anyone attempting this position with a short molecule will meet the same constraint. It does not close either editing route, because the residual risk is concentrated at one named place rather than scattered, and a concentrated risk at a named locus is a sequencing target rather than an unknown hazard. It does mean that place has to be checked in any experiment.
Two editors, and the trade between them
